server plus

CompTIA Server+ Certification Support Skills (2009 Objectives)

Study Notes

G601eng ver021

Tel: +44 (0)20 7887 7999 Fax: +44 (0)20 7887 7988 Email: [email protected]

Acknowledgements

www.gtslearning.com

Course Developer ......................................................... gtslearning

Editor ..................................................................... James Pengelly

This courseware is owned, published, and distributed by gtslearning, the world’s only specialist supplier of CompTIA learning solutions. [email protected] +44 (0)20 7887 7999 +44 (0)20 7887 7988 Three Elysium Gate, 126-128 New Kings Road, London, SW6 4LZ, UK

COPYRIGHT

This courseware is copyrighted © 2010 gtslearning. Product images are the copyright of the vendor or

manufacturer named in the caption and used by permission. No part of this courseware or any training material

supplied by the publisher to accompany the courseware may be copied, photocopied, reproduced, or re-used in

any form or by any means without permission in writing from the publisher. Violation of these laws will lead to

prosecution.

All trademarks, service marks, products, or services are trademarks or registered trademarks of their respective

holders and are acknowledged by the publisher.

LIMITATION OF LIABILITY

Every effort has been made to ensure complete and accurate information concerning the material presented in

this course. Neither the publisher nor its agents can be held legally responsible for any mistakes in printing or for

faulty instructions contained within this course. The publisher appreciates receiving notice of any errors or

misprints.

Information in this course is subject to change without notice. Companies, names, and data used in examples

herein are fictitious unless otherwise noted.

Where the course and all materials supplied for training are designed to familiarise the user with the operation of

software programs and computer devices, the publisher urges the user to review the manuals provided by the

product vendor regarding specific questions as to operation.

There are no warranties, expressed or implied, including warranties of merchantability or fitness for a particular

purpose, made with respect to the materials or any information provided herein. Neither the author nor publisher

shall be liable for any direct, indirect, special, incidental, or consequential damages arising out of the use or the

inability to use the contents of this course.

ISBN

ISBN: 978-1-84005-299-2

Warning All gtslearning products are supplied on the basis of a single copy of a course per student.

Additional resources that may be made available from gtslearning may only be used in conjunction with courses

sold by gtslearning. No material changes to these resources are permitted without express written permission

from gtslearning. These resources may not be used in conjunction with content from any other supplier.

If you suspect that this course has been copied or distributed illegally,

please telephone or email gtslearning.

CompTIA Server+ Certification Support Skills (2009 Objectives) Table of Contents

Page iii

Table of Contents

Course Introduction i

Table of Contents ............................................................................................................. iii About This Course ........................................................................................................... vii

Module 1 / Planning Servers 1

Module 1 / Unit 1 Server Fundamentals 3

Servers as Hardware and Software ................................................................................... 3 Server Form Factors .......................................................................................................... 8 Racks .............................................................................................................................. 10

Module 1 / Unit 2 Configuration Management 17

Project Planning Concepts .............................................................................................. 17 Server Pre-installation Planning....................................................................................... 20 Server Upgrade Plans ..................................................................................................... 24 Implementing the Plan ..................................................................................................... 26 Verifying the Plan ............................................................................................................ 27 Configuration Management ............................................................................................. 28 Documentation ................................................................................................................ 31 Equipment Disposal ......................................................................................................... 37

Module 1 / Unit 3 Storage and RAID 39

Hard Drives ..................................................................................................................... 39 The SCSI Interface .......................................................................................................... 42 Serial Attached SCSI (SAS) ............................................................................................ 52 The Serial ATA Interface ................................................................................................. 54 Drive Arrays (RAID) ......................................................................................................... 56

Module 1 / Unit 4 Installing an NOS 69

Installation Procedures .................................................................................................... 69 File Systems .................................................................................................................... 73 Installation Methods ......................................................................................................... 78 Configuring the NOS ....................................................................................................... 82 Performing Shut Down .................................................................................................... 90

Module 1 / Summary Planning Servers 97

Course Introduction CompTIA Server+ Certification Support Skills (2009 Objectives)

Page iv

Module 2 / Configuring Servers 99

Module 2 / Unit 1 Server Networking 101

Ethernet ......................................................................................................................... 101 TCP/IP ........................................................................................................................... 106 Virtual LANs (VLAN) ...................................................................................................... 111

Module 2 / Unit 2 Directory Services 114

Configuring Directory Services ...................................................................................... 114 Managing Users ............................................................................................................ 116

Module 2 / Unit 3 File and Print Services 126

Configuring File and Print Services ................................................................................ 126 External Storage Technologies ...................................................................................... 136 Disk and Volume Management ...................................................................................... 140

Module 2 / Unit 4 Site Security 148

Designing a Secure Site ................................................................................................ 148 Site Security Controls .................................................................................................... 152

Module 2 / Unit 5 Server Security 159

Secure Network Topologies ........................................................................................... 159 Firewalls and Proxy Servers .......................................................................................... 163 Malware Protection Software ......................................................................................... 168

Module 2 / Summary Configuring Servers 173

CompTIA Server+ Certification Support Skills (2009 Objectives) Table of Contents

Page v

Module 3 / Application Servers 175

Module 3 / Unit 1 Configuring Server Roles 176

Installing Roles and Applications ................................................................................... 176 Network Configuration and Access ................................................................................ 179 Database and Application Servers ................................................................................. 189 Web and FTP Servers ................................................................................................... 191 Messaging Servers ........................................................................................................ 197

Module 3 / Unit 2 Virtualisation Technologies 199

Virtualisation Defined ..................................................................................................... 199 Virtual Platform Applications .......................................................................................... 202 Virtualisation Best Practices .......................................................................................... 205

Module 3 / Summary Application Servers 211

Module 4 / Managing and Upgrading Servers 213

Module 4 / Unit 1 Management and Monitoring Tools 215

Developing a Server Management Plan ........................................................................ 215 Network Management Protocols .................................................................................... 216 Server Access Tools ...................................................................................................... 221 Server Monitoring .......................................................................................................... 226 System Logs.................................................................................................................. 238

Module 4 / Unit 2 Environment and Maintenance 242

Preventative Maintenance ............................................................................................. 242 Power ............................................................................................................................ 243 Server Environment ....................................................................................................... 257

Module 4 / Unit 3 Installing and Upgrading Hardware 265

Adding and Removing Components .............................................................................. 265 Accessing the Case ....................................................................................................... 266 CPU .............................................................................................................................. 272 Memory ......................................................................................................................... 280 Expansion Cards ........................................................................................................... 289 Firmware ....................................................................................................................... 293

Module 4 / Summary Managing and Upgrading Servers 300


Page vi

Module 5 / Troubleshooting Servers 301

Module 5 / Unit 1 Troubleshooting Procedures 303

Troubleshooting Models and Processes ........................................................................ 303 Troubleshooting Steps ................................................................................................... 305 Approaching Troubleshooting ........................................................................................ 315

Module 5 / Unit 2 Troubleshooting Scenarios 317

Troubleshooting Hardware Problems ............................................................................ 317 Troubleshooting Storage ............................................................................................... 324 Troubleshooting Network Problems ............................................................................... 331 Troubleshooting Software Problems .............................................................................. 342

Module 5 / Unit 3 Disaster Recovery 348

Disaster Recovery Planning .......................................................................................... 348 Fault Tolerance and Redundancy .................................................................................. 350 Health and Safety .......................................................................................................... 355 Recovery ....................................................................................................................... 359 Backup Strategies ......................................................................................................... 362

Module 5 / Summary Troubleshooting and Disaster Recovery 376

Index 377

CompTIA Server+ Certification Support Skills (2009 Objectives) About This Course

Page vii

About This Course

Who Should Follow This Course?

This course is intended for students wishing to qualify with CompTIA Server+ Certification for advanced level technical competency of server issues and technology. It is also suitable for PC support technicians wanting to improve their skills in support and administration.

The course has been developed to produce server support technicians who are capable of taking and passing CompTIA's Server+ Certification exam or other similar qualifications.

What are the Course Prerequisites?

Ideally, you should have successfully completed CompTIA A+ Certification and have around 12 months' experience of PC support. It is not necessary that you pass the A+ exams before completing Server+ certification, but this is recommended.

Regardless of whether you have passed A+, it is recommended that you have the following skills and knowledge before starting this course:

■ Know the function and basic features of the components of a PC.

■ Know the PC startup process.

■ Use Windows to create and manage files and use basic administrative features (Explorer, Control Panel, Management Consoles).

■ Basic network terminology (such as OSI Model, Topology, Ethernet, TCP/IP).

Optionally, you can take a prerequisites test to check that you have the knowledge required to study this course at www.gtssupport.com/flower27/SK0-003/index.htm.

Course Outcomes

The main aim of this course is to help to prepare you for CompTIA's Server+ Certification exam. Server+ Certification is internationally recognised by many corporations. Indeed, CompTIA Server+ Certification is a prerequisite qualification for employment (and is endorsed) by many leading computer manufacturers and vendors on a global basis.

This course will teach you advanced level technical competency of server issues and technology, including installation, configuration, upgrading, maintenance, environment, troubleshooting and disaster recovery.


Page viii

Study of the course can also help to prepare you for other, similar technical support qualifications and act as a groundwork for more advanced training. Other qualifications available include:

■ CompTIA Network+ - a foundation-level certification of competency in network installation and configuration.

■ Cisco Certified Network Associate (CCNA) - a foundation-level certification of competency in Cisco networking appliance installation and configuration.

■ Microsoft Certified Systems Administrator / Engineer / IT Professional (MCSA / MSCE / MCITP) - Windows-specific qualifications; passing CompTIA’s Network+ Certification plus either Server+ or Security+ Certification can satisfy the requirements for the elective part of the certification, as well as providing a solid groundwork for the further study required for MCSA / MCSE (Windows Server 2003 track) or MCITP (Windows Server 2008 Server Administrator / Enterprise Administrator track).

■ Help Desk Support Analyst - The Help Desk Analyst certification series, administered by the Help Desk Institute (www.thinkhdi.com), certifies learners’ customer service and Help Desk management skills. Various levels of certification are available, including Customer Support Specialist, Help Desk Analyst and Help Desk Manager.

About the Course Material

The course material has been prepared as an aid for your use throughout the training course. You may keep this manual for your own reference after the course is finished. We hope you will find the course material useful for future reference.

Course Organisation

This course book contains the study notes for you to refer to in class and to review at home as you prepare for the exam. The course is divided into several modules, each covering a different subject area. Each module is split into a series of units containing related topics for study. Each unit has a set of review questions designed to test your knowledge of the topics covered in the unit. Answers to the review questions are located in the companion volume.

Throughout the course, there will be ample opportunity for you to learn through practical work. A series of 'hands-on' labs help to familiarise you with the concepts and technologies that are taught on this course.

At the back of the book there is an index to help you look up key terms and concepts from the course.


Page ix

The accompanying book contains a list of the CompTIA certification objectives (and where in the study notes you can find useful material to prepare for each objective), tips for taking the CompTIA exams, the practical labs for you to complete in class, a glossary of terms and concepts used in computer support, and answers to the end of unit review questions.

When you have completed the course and want to prepare for the exam, you can take a practice test at www.gtssupport.com/flower27/SK0-003/index.htm.

Conventions Used in the Course

Certain conventions have been followed to help you use this course material. These are especially useful for following the practical lab exercises.

Bullet Points

Steps for you to follow in the course of completing a task or hands-on exercise and review questions are indicated by numbered bullet points. Other bullet points indicate learning objectives and feature lists.

File Conventions

The steps to follow to open a file or activate a command are shown in bold with arrows. For example, if you need to access the Control Panel in Windows, this would be shown in the text by:

Example: Select Start > Settings > Control Panel

Text Conventions

Commands

Commands or information that needs to be supplied by you that are entered from the keyboard are shown in Courier New bold.

Example: Type [email protected]

Displayed text and buttons

Information that is displayed on the screen by the computer is shown in sans serif bold. This includes button text and messages.

Examples: Click OK, Click Continue...


Page x

Key to Symbols Used in the Notes

Icon Meaning

A note or warning about a feature.

More information on this topic can be found in the section mentioned.

An area for your notes.

Review questions to help test what you have learned.

A hands-on exercise for you to practise skills learned during the lesson.

CompTIA Authorised Quality Curriculum Program

The logo of the CompTIA Authorised Quality Curriculum (CAQC) program and the status of this or other training material as "Authorised" under the CompTIA Authorised Quality Curriculum program signifies that, in CompTIA's opinion, such training material covers the content of CompTIA's related certification exam.

The contents of this training material were created for the CompTIA Server+ Certification exam (exam code: SK0-003) covering CompTIA certification objectives that were current as of April 2009.

CompTIA has not reviewed or approved the accuracy of the contents of this training material and specifically disclaims any warranties of merchantability or fitness for a particular purpose. CompTIA makes no guarantee concerning the success of persons using any such “Authorised” or other training material in order to prepare for any CompTIA certification exam.


Page xi

How to Become CompTIA Certified

This training material can help you prepare for and pass a related CompTIA certification exam or exams. In order to achieve CompTIA certification, you must register for and pass a CompTIA certification exam or exams.

In order to become CompTIA certified, you must:

1) Select a certification exam provider. For more information please visit certification.comptia.org/resources/registration.aspx

2) Register for and schedule a time to take the CompTIA certification exam(s) at a convenient location.

3) Read and sign the Candidate Agreement, which will be presented at the time of the exam(s). The text of the Candidate Agreement can be found at certification.comptia.org/resources/candidate_agreement.aspx

4) Take and pass the CompTIA certification exam(s).

For more information about CompTIA’s certifications, such as their industry acceptance, benefits, or program news, please visit certification.comptia.org

CompTIA is a not-for-profit information technology (IT) trade association. CompTIA’s certifications are designed by subject matter experts from across the IT industry. Each CompTIA certification is vendor-neutral, covers multiple technologies, and requires demonstration of skills and knowledge widely sought after by the IT industry.

To contact CompTIA with any questions or comments, please call (1) (630) 678 8300 or email [email protected].

It is CompTIA's policy to update the exam regularly with new test items to deter fraud and for compliance with ISO standards. The exam objectives may therefore describe the current "Edition" of the exam with a date different to that above. Please note that this training material remains valid for the stated exam code, regardless of the exam edition. For more information, please check the FAQs on CompTIA's website (certification.comptia.org/customer_service).


Page xii

This page left blank intentionally.

No part of these notes may be reproduced in any form, electronic or printed, without the written permission of a director of gtslearning International Limited. If you suspect that these notes have been unlawfully copied, please telephone +44 (0)207 887 7999 or email [email protected]

Study Notes Planning Servers

Page 1

Module 1 / Planning Servers

In this module, you will learn about server form factors and storage technologies. You will also learn how to plan and specify servers and document the server environment and how to install or deploy a Network Operating System (NOS).

The following table lists the CompTIA Server+ domain objectives and content examples covered in the units in this module:

Unit Domain Objectives / Examples 1.1 Server

Fundamentals 1.2 Deploy different chassis types and the appropriate components Form Factor [tower, rack, blade] (Space utilisation ([U size, height, width, depth]) • Power buttons • Reset buttons • Diagnostic LEDs

2.4 Explain different server roles, their purpose and how they interact Explain the difference between a workstation, desktop and a server

1.2 Configuration Management

4.1 Write, utilise and maintain documentation, diagrams and procedures Follow pre-installation plan when building or upgrading servers • Labelling • Diagram server racks and environment topologies • Hardware and software upgrade, installation, configuration, server role and repair logs • Document server baseline (before and after service) • Original hardware configuration, service tags, asset management and warranty • Vendor specific documentation (Reference proper manuals, Websites, Support channels [list of vendors])

4.2 Given a scenario, explain the purpose of the following industry best practices Follow vendor specific server best practices (Documentation, Tools, Websites) • Explore ramifications before implementing change – determine organisational impact • Communicate with stakeholders before taking action and upon completion of action • Comply with all local laws / regulations, industry and corporate regulations • Purpose of Service Level Agreement (SLAs) • Follow change control procedures • Equipment disposal

4.5 Given a scenario, classify physical security measures for a server location Secure documentation related to servers (Passwords, System configurations, Logs)

No part of these notes may be reproduced in any form, electronic or printed, without the written permission of a director of gtslearning International Limited. If you suspect that these notes have been unlawfully copied,

please telephone +44 (0)207 887 7999 or email [email protected]

Module 1 / Unit 1 CompTIA Server+ Certification Support Skills (2009 Objectives)

Page 2

Unit Domain Objectives / Examples 1.3 Storage

and RAID1.3 Differentiate between memory features / types and given a scenario select appropriate memory RAID and hot spares 1.6 Given a scenario, install appropriate expansion cards into a server while taking fault tolerance into consideration Storage controller [SCSI, SATA, RAID] (SCSI low voltage / high voltage [LVD/HVD], SCSI IDs, Cables and connectors, Active vs. passive termination) 3.1 Describe RAID technologies and its features and benefits Hot spare • Software vs. hardware • Cache read/write levels (data loss potential) • Performance benefits and tradeoffs 3.2 Given a scenario, select the appropriate RAID level 0, 1, 3, 5, 6, 10, 50 • Performance benefits and tradeoffs 3.3 Install and configure different internal storage technologies Hot swappable vs. non-hot swappable • SCSI, Ultra SCSI, Ultra320 (termination), LUNs • SAS, SATA • Flash • Controller (firmware levels) • Hard drive (firmware, JBOD) 6.5 Given a scenario, effectively troubleshoot storage problems, selecting the appropriate tools and methods Storage tools (RAID array management, Array management)

1.4 Installing an NOS

1.4 Explain the importance of a Hardware Compatibility List (HCL) Vendor standards for hardware • Memory and processor compatibility • Expansion cards compatibility 2.1 Install, deploy, configure and update NOS (Windows / *nix) Installation methods (optical media, USB, network share, PXE) o Imaging – system cloning and deployment (Ghost, RIS/WDS, Altiris) • Bootloader • File systems (FAT, FAT32, NTFS, VMFS, ZFS, EXT3) • Driver installation (Driver acquisition, Installation methods, Require media) • Configure NOS (Device, OS environmental settings) • Patch management 2.4 Explain different server roles, their purpose and how they interact Application server (Update server) • Server shut down and start up sequence (one server vs. multiple servers vs. attached components)


Study Notes Server Fundamentals

Page 3

Module 1 / Unit 1 Server Fundamentals

Objectives

On completion of this unit, you will be able to:

■ Distinguish between servers and desktops and workstations and know what is meant by fault tolerance, availability, scalability, and load balancing.

■ Identify basic server software functions and roles.

■ Describe the features, advantages and disadvantages of tower, rack, and blade servers.

■ Describe the features of rack-mounted equipment.

■ Identify commonly-used rack-mounted peripherals and accessories.

■ Identify the external features of a server case, such as the power button and diagnostic LEDs.

Servers as Hardware and Software

A server is any computer providing services to other computers, but usually the term server implies a powerful computer that supports a number of users simultaneously (in a computer network).

The term "server" also means a software application that provides services to client applications. You will learn about examples of server applications later in the course.

Servers and Desktops

There is a distinction between server-class hardware and computers used as desktop PCs ("desktops") and workstations. A desktop PC is a low or medium spec machine provided to an end-user1.

1 "Desktop" in this sense doesn't refer to the desktop form factor case. Most home and office computers come in tower cases but can still be referred to as desktops.




Page 4

Any desktop PC can act as a server, but when a computer is designed to act as a server computer, it will have a number of features that distinguish it from a desktop computer.

Some of the major manufacturers of PC servers include Hewlett-Packard (HP), Dell, Lenovo, IBM, Sun Microsystems, Fujitsu-Siemens, and Apple.

High Availability and Fault Tolerance

It is a well-known maxim that time costs money. Any time that an organisation's staff spend in the workplace without access to the computer system (downtime) is likely to be time wasted, and therefore a significant cost. If you calculate this cost based on the amount of time lost multiplied by employees' salaries, you can provide a cost-justification for upgrading the server equipment to improve reliability.

Availability is not restricted to catastrophic crashes or complete system failures. A system that is slow is also costing time (and probably contributing to stress and low morale in the workplace).

Availability is also critical when providing an interface for customers. If a web server crashes or responds slowly, it makes the company look inefficient at best; at worst it makes the company invisible (a customer can switch to a competitor's site with a couple of mouse clicks).

Fault tolerance refers to a server's ability to keep serving even if something goes wrong.

Availability is provided through the robustness of computer components and by providing for redundancy in the system (additional components that take over in the event of failure or extra capacity that can be utilised in high load conditions). Where redundancy is provided, the component is said to be n+1, where n is the number of components in use and 1 is a spare to be used in the event of failure.

Another important feature is that as many configuration, maintenance, and repair operations as possible can take place without interrupting services (hot-swappable components for instance).

Load Balancing

Load balancing refers to distributing requests between multiple components. From the point-of-view of the client, there is only one resource.



Page 5

In some configurations, having multiple components can provide load balancing and redundancy. If one of the components fails, the others can continue to provide the service, albeit at a reduced level of performance. Other load balancing configurations do not provide redundancy; if one component fails, the whole system goes down.

Scalability

Scalability means the ability to improve the performance of a server without diminishing its usefulness or increasing its cost disproportionately. A well-designed server is able to expand to meet the increased needs of its users. For example, if a company doubles its staff, it is likely to require about twice as much capacity from its server.

If components such as an additional processor and more memory and disk drives can be added to the existing server then the server could be said to be scalable. The system is no more difficult to manage than before and performing the upgrade will have entailed a minimum of downtime.

Contrast this with the circumstance in which a company had to buy a new server or an additional server. These solutions cannot be said to be scalable as they involve greater expense and complexity of administration.

Workstations

A workstation is a high-spec machine provided to "power users", such as designers or developers. A workstation may have some of the features associated with servers, such as multiple processors and redundant disks (RAID). A workstation may also be able to function quite competently as a server in a small workgroup (up to about 10 users)2.

At the entry-level, server computers are not that much more expensive than desktops. Most businesses would require enterprise-class features however and these quickly increase the required budget.

2 Workstations can also be provided through server-like infrastructure. For example, HP produce blade workstations accessed remotely via client PCs. The main difference between such a workstation and a server is that the workstation is likely to have a high-performance graphics adapter. Servers only require basic graphics functionality while workstations used in the design industry need to be able to do 3D and texture processing.




Page 6

Server Roles

While "server" can refer to the computer hardware, the term also refers to software. This can broadly be divided into Network Operating System server platform and server applications software.

Server Platform

The server platform is the underlying Network Operating System (NOS). This will usually be a version of Windows Server or some "flavour" of Linux.

The server platform is likely to include a number of basic roles enabling the server to provide local and remote user authentication, network services (routing, DHCP, and DNS), file and printer sharing, and web, FTP, and email.

The current trend is for virtualised server platforms. This means that a single hardware server, installed with a virtual machine manager or hypervisor, hosts multiple OS platforms. These multiple instances or VMs could all be running similar services (web servers for instance) or different types of application.

Web Hosting

Running web services is one of the most common server roles. As mentioned above, any server platform can host a basic website but web services are very often provisioned on a larger scale.

Windows Server Internet Information Services web server



Page 7

Internet Service Providers (ISPs) operate huge web server farms to provide web hosting services to subscribers. Enterprise websites are also backed up by multiple servers for load balancing and availability.

Messaging, Groupware, and Collaboration

Messaging servers provide services such as email, Voice over IP (VoIP), Instant Messaging (IM), and web conferencing. The same machines may also run "groupware" services, allowing users to arrange meetings and collaborate on projects and documents.

Media / Streaming

Servers designed to serve audio and video files to large numbers of users are often provisioned as dedicated hardware, separated from the web server(s) to improve performance and availability.

Application, Database, and Middleware

A client-server application can be designed for any sort of business function (Customer Relationship Management [CRM] and accounting are typical examples). These applications often feature split front- and back-end servers. The front-end servers host the user interface while the back-end servers host the database. There may also be "middle" tiers supporting functions such as business logic and data queries.

Small Business

Enterprise networks run multiple server applications on multiple server machines, physical and virtual. Small and Medium Size Enterprises (SME) may use "one-box" solutions, such as Microsoft's Small Business Server. As well as the underlying OS platform, the server provides messaging and calendaring (Exchange), collaboration (SharePoint), database (SQL Server), and (in some editions) a firewall and proxy server (Internet Security and Acceleration [ISA] Server).




Page 8

Server Form Factors

A server's form factor is the type of case it uses.

Tower Servers

A tower server uses the same sort of case as a PC. Often the case is substantially larger and the design gives more thought to providing airflow through the case, improving cooling. This allows flexibility in terms of placement and is the cheapest option.

Tower servers are available in a wide range of models, allowing for different budgets and requirements. They are also usually the easiest machines to customise and upgrade.

HP full tower server

A tower server is a good option if only a few servers need to be stored. However, when the server equipment needs to be integrated with network equipment or where an organisation needs to be able to grow processing power rapidly and effectively, a rack-mounted solution is required.

Rack-mounted Servers

Rack servers are designed to bolt into racks. They are built to fit within a standard width (19" / 48.26cm) and height. The height of the case will be a multiple of 1.75" / 4.45cm called a U.

They look somewhat similar to the old desktop form factor PCs, but are not designed to be placed free-standing on a desk.

HP Proliant 4U (left) and 1U (right) rack-mounted servers

Rack servers can be far more compact than tower servers, but contain the same components. Notably, each server has its own cooling solution.



Page 9

The cost of buying rack equipment makes rack-mounted servers seem a more expensive option than tower servers, but these costs can be offset by the greater security and manageability of rack-mounted equipment. Rack-mounted equipment also allows the server array to grow more easily.

Racks are said to offer density, as units can be stacked vertically as well as horizontally.

Blade Servers

A blade server is a space-saving variation of a rack server. The power and cooling systems are not incorporated within the server unit. Instead, servers are housed in very thin cases installed together in an enclosure like books in a book case. Each enclosure shares a common power and cooling system and also I/O and network ports.

HP Proliant blade server (front view) and enclosure (back view)

Each machine is typically identically configured in terms of number and type of processors, system memory, and disk drive provision.

Blade servers offer the best density of any server form factor. A rack that can fit 42 1U servers could fit 60 blade servers. The individual server units are usually fairly low cost (though not as cheap as similarly specified 1U rack servers), allowing the organisation to grow capacity organically.

An extra consideration is the overall power consumption of the rack, especially as energy prices are likely to continue to rise. With shared power and cooling, blade servers are typically more efficient than ordinary rack-mounted servers. Also, cable management is significantly easier.

As a relatively new technology, one problem with blade servers is the lack of standards. Machines from different manufacturers (IBM, HP, and Dell) cannot be integrated in the same chassis.




Page 10

The management software offered with the solution also needs to be evaluated carefully, as it has a strong impact on the usability of the server array. Blade servers are ideal for supporting server clusters. Often servers can be hot-swapped out of the array, allowing for fault tolerance and redundancy.

Organisations considering a solution would need to make a careful comparison of the costs and benefits of blade servers versus the 1U equivalent.

Racks

A rack is a specially configured steel shelving system for server and network equipment. Using a rack allows equipment to be stored more securely and compactly than ordinary desks or shelving would allow for.

Rack Specifications

HP rack for blade server enclosures

Rack or blade servers and network equipment are EIA standard width (19" / 48.26cm), allowing them to be screwed into the rack directly. Non-standard components, such as a tower server or monitor, can be installed on shelves.

Racks can be purchased with varying depths (from 30-40" / 76.2-101.6cm) and heights. Rack height is measured in EIA units of 1.75" / 4.45cm called U. Racks are sold in heights from 8U to 48U. Rack-compatible equipment is designed with a vertical height quoted in U so you can plan exactly how much vertical space you require.



Page 11

Racks have a maximum weight capacity. Rack accessories such as shelves will also have a maximum weight capacity.

Most racks are designed to be free-standing, though smaller wall-mounted cabinet units are also available. Free-standing racks can be bolted together in rows.

It is possible to obtain racks that are dust and water-resistant, which can be useful for highly-sensitive equipment or where there is the risk of flood (though such racks are water-resistant, not water-proof).

Racks can be purchased direct from the major server vendors or from third parties. Third-party racks are sometimes designed for a specific vendor's equipment. Many vendors have online specification tools which you can use to build a complete network solution (including the important question of what colour it should be).

Designing rack layout in Microsoft Visio

Rack Accessories

A number of accessories are available to make the basic rack configuration fit for purpose.

Shelving and rails

Most equipment screws directly into the rack. Devices are often mounted on rails so that they can be slid out of the rack for configuration. Other equipment can be mounted on shelves, sliding or otherwise.




Page 12

Doors and panels

Side panels and banking plates (to cover unused slots) are used to improve air flow. Lockable doors (front and rear) obviously help to prevent access to the equipment.

Server cooling works on the basis of cold air through the front and warm air out the back. Energy is wasted if the warm air is allowed to contaminate the cool air. See Unit 4.2 for more information about environment and cooling.

Stabilisers

Some racks can be wheel mounted; others require stabilisation or levelling. It is also possible to add ballast; weight to balance a rack where the weight of devices in the rack cannot be distributed evenly.

KVM (Keyboard, Video, Mouse)

HP rack with KVM

Rack mounted KVM usually consists of a drawer with LCD screen, keyboard, and trackball. KVM units allow the control of multiple servers with a single set of input/output devices. KVMs have a number of ports corresponding to the number of servers they can control. A switch or keyboard combination is used to switch between servers.

KVM are cabled using standard connectors (PS/2 or USB for the mouse and keyboard and VGA, DVI, or HDMI for the video signal). Some KVM also support remote or dial-up access (KVM over IP).



Page 13

Other accessories

A variety of other accessories are available, including cable management, combination locks for doors, power strips, fans, and monitoring equipment.

Rack Installation

The first step is to place the racks in the correct location in the server room. Multiple rows should be placed back-to-back not front-to-back to maximise cooling and there should be about 3' (1 meter) clearance (for service access and air flow).

If necessary, bolt the racks together and check that they are stable, adjusting levelling bolts if necessary.

Generally speaking, large and heavy items should be placed lower in the rack. Weight distribution is probably the main factor in determining where an item should be placed in the rack, but also consider the following:

■ Try not to place backup tape drives and network cabling near UPS, which can generate EMI3 (though the risk of data corruption is low, except with the most powerful UPS).

■ A KVM will obviously need to be positioned at standing height.

Use a numbering system to identify the servers to KVM port switch buttons and position the servers in number order (top-to-bottom and left-to-right). Incorporate the number into the server's desktop wallpaper so that it is easy to see which server is being controlled at any one time.

■ Check cable lengths and plan ahead for any devices that need to be connected, as removing and re-inserting equipment is time-consuming. Keep cables tidy using brackets and ties. Do not allow cables to hang in front of devices.

■ Check that equipment on pull-out rails does not cause the rack to tip. If there is any sign of instability, add ballast to compensate. Also check that pulling the tray out does not dislodge any connectors. Leave enough slack in the cable for an item to be pulled out along the length of its rail.

Never pull out more than one item at a time. 3 Electromagnetic Interference (EMI) is interference in one electromagnetic device caused by the radiation generated by another.




Page 14

Power and Reset Buttons

One of the key distinctions between a desktop PC and a server is that you never turn a server off without a very good reason and a lot of planning!

Consequently, the power button on a server will often be protected by some sort of locking mechanism. Even if there is no lock, it is usually the case that you need to hold down the power button for a few seconds for anything to happen.

Often the power button may not completely power down the server but put it into standby mode instead (allowing it to be woken by remote management software) and you may need to remove the AC cord to completely power off the server.

Some servers may also feature a reset button, which will shut down then reboot the server. This is usually only a feature of older server models though.

The precise operation of server-level hardware is very often vendor- or model-specific. Check the user or maintenance guide carefully if you are at all unfamiliar with the server model.

Diagnostic LEDs

Servers can be managed remotely, with alerts and event information being sent to a management station. Sometimes however, there is no substitute to patrolling the data centre floor looking for problems. Server vendors make this sort of frontline troubleshooting and maintenance easier by providing a number of status and diagnostic LEDs on the front and back of the chassis.

For example, the HP DL 165 server provides the following diagnostics:

HP DL 165 rack-mounted server with 1) UID, power, health, and NIC activity LEDs and

2) Drive LEDs

1

2



Page 15

■ Power - the LED is green for on and amber for standby.

■ System health - this shows red when the server is shutting down or in a critical state, amber when it is powering up or in a degraded state, and green otherwise.

■ UID (Unique ID) - this blue LED can be activated remotely and helps a technician locate the server in the rack. It flashes when there is a remote management connection. Remember to switch the UID off when you have finished whatever work you needed to do.

■ NIC(s) - steady green represents a network link and flickers represent activity on a network card.

■ Hard drive online - this LED can be set to blue using management software (to ID the drive) or shows orange if the drive has failed.

■ Hard drive activity - flashes to show disk activity.

■ Optical disc - flashes to show disc activity.

The rear of the chassis has a rear UID indicator and NIC LEDs to show the link status, speed, and activity.

HP DL 165 rack-mounted server with 1) Rear UID indicator and 2) NIC speed and

activity LEDs

Some servers may come with small LCDs to display status messages (such as "Power supply failed") on the front panel.

12

2




Page 16

Review Questions Module 1 / Unit 1 / Server Fundamentals

Answer the following questions. The correct answers are in the "Labs and References" volume.

1) Apart from fault tolerance and scalability, what general feature distinguishes server-class systems from desktop PCs? Load balancing - distributing requests between multiple components.

2) What is the advantage of rack-mounted or blade servers over the tower form factor? Better density (fitting more servers into the same space).

3) You are installing multiple racks in a server room. What layout guidelines should you try to follow? Allow a 3 foot service gap behind each rack (for access and cooling). Install racks back-to-back not back-to-front for optimum cooling.

4) When installing equipment into racks, what general guidelines should you try to follow? Check cables will run between devices that need to be connected. Position heavier devices lower in the rack. Check pull-out trays do not cause tipping and add ballast if necessary. Put servers controlled by KVM in a logical order.

5) Why might pressing the power button on a server have no effect? The button may be locked through hardware or software to prevent unauthorised shutdowns.

6) What might be the significance of a blue LED on the back of a server chassis? This is most likely to be a UID to help identify the server picked for some maintenance task out of the others in the rack.


Study Notes Configuration Management

Page 17

Module 1 / Unit 2 Configuration Management

Objectives


■ Describe key project planning concepts.

■ Identify the main considerations to make when installing or upgrading servers.

■ Understand the role of configuration management policies, procedures, and documentation in server administration.

■ Understand the use of baseline configurations and performance criteria.

■ Compare and update server and rack schematics, diagrams, and job logs.

Project Planning Concepts

In order to make the rollout of new equipment or software or the upgrade of existing systems successful, the operation must be carefully planned and implemented. Effective project planning depends on understanding and implementing some critical concepts, as described briefly below.

The characteristics of a project are that it has a goal, schedule, budget, and resources. The project should be controlled by the project manager, who is accountable for the project's success.

Defining the Project

The first step in project planning is to define the project clearly. This involves the following considerations:

■ State why the project is necessary.

■ State the goal of the project.

■ State the objectives that will be met by completing the project.

■ Identify the scope of the project.




Page 18

■ Make an initial estimate of the timeframe and resources that will be required.

■ Identify risks that could damage the chance of success.

What is the project for? (Business needs and benefits)

A project should be driven by precise business needs and deliver measurable business benefits. It is important that the benefits be measurable so that there are clear criteria for success or failure.

Consider what is wrong with the IT Manager telling the Financial Director that a new server is required because it will "make things more efficient and stop all those problems we've been having". The Financial Director would be correct in firing a number of questions back:

■ What is the cause of the existing problems? Why is a new server purchase the best way of resolving them?

■ How will efficiency improve? Will it lead to more sales, greater customer satisfaction, faster product development? Can you quantify the improvement?

Goals and objectives

A goal is a clear statement of intent, similar to a mission statement. For example:

■ "Implement an e-commerce portal to market and sell products directly to customers over the web."

Objectives are the things that will bring you towards your goal. Well-defined objectives are often described as SMART:

■ Specific - the objective is stated unambiguously in terms that everyone connected with the project understands.

■ Measurable - there is an unambiguous means of deciding whether the objective has been achieved.

■ Achievable - the objective can be realised using the resources provided.

■ Realistic - the objective is within the scope of other objectives and is tied into business needs and benefits.

■ Timely - the objective can be completed within an appropriate timeframe.



Page 19

With an IT system, it is also typical to complete a requirements analysis. This means identifying all the things the planned system should do and what its operational parameters will be. Requirements could include business, technical, and legal (compliance) factors. Analysis would proceed by examining existing systems, questioning users and management, looking at competitors, and so on.

Taken together, the objectives, requirements, timeframe and resources define the scope of the project (that is, what things the project will try to accomplish). It is important to realise the adverse impact changing any of these factors will have once the project is up and running (the difficulty of hitting "moving goalposts").

Sponsors and Stakeholders

Once the project outline has been drafted, the next step is to get the commitment to put the project into action.

The project sponsor is the senior manager or director within the organisation with the closest interest in the project. The sponsor will generally authorise the budget and resources and sign off on project completion (or call a halt to a failing project).

Stakeholders are those in the organisation who could be impacted by the project or could influence its outcome. It is important to establish a commitment for the project to succeed among stakeholders (and senior management in general).

Planning the Project

When the project is authorised, the project manager (the person accountable for the project's success) will create a project plan. This will consist of tasks and deliverables that fulfil the objectives of the project. The point of having a plan is to show responsibility for completing tasks and deliverables clearly and to monitor progress on each task.

Microsoft Project planning tool




Page 20

Project plans are often created using tools such as Microsoft Project. These help the project manager to identify task dependencies, schedule deliverables, and allocate resources.

Implementing the Project

During the implementation of the project, it is vitally important to establish effective communication channels within the project team and to the project sponsor and stakeholders.

The project manager must monitor progress achieved on tasks and deal with any delays. In a complex project, the manager also needs to be able to motivate and manage the project team.

Closing the Project

When the project is closed (either because it has achieved its objectives or because it has been stopped), it is normal to review what was done against what was promised. The goal is to work out how effective the project was and if mistakes were made, to learn from them.

Server Pre-installation Planning

Planning for the installation of a new server requires three main tasks:

■ Establish what specification is required. How powerful does the server need to be? What components are to be installed?

■ Establish how the server will be integrated with existing hardware, software, and network resources.

■ Create an ongoing management plan. Who will be responsible for maintaining the server? Is there current expertise within the organisation to support the chosen NOS or applications software? Will users require retraining or will operations management procedures require updating?

Specifying the Server

The first job in specifying a new server is to know as far as possible what it is going to do, both now and in the future. Remember from the notes on project planning that purchasing decisions should be business-led, arising from a specific business need and offering demonstrable benefits (such as a percentage increase in sales or call throughput).



Page 21

Given that, it is also true to say that spending the least amount of money possible for a minimum spec machine is not necessarily cost-effective. Installing a new server is not a completely straightforward task, and upgrading one can be even more complex, so future-proofing is likely to make sense economically, but this benefit needs to be quantified to justify the additional up-front cost.

If the requirements and objectives are clearly specified and related to concrete business benefits, the budget is far more likely to be approved.

In terms of selecting components, remember that reliability, security, and availability of technical support should be given greater consideration than pure performance. New and unproven technologies are significant risk factors. You should read up on the components used in the server, looking for reviews, white papers and technical tips on the Internet. You also need to evaluate the vendor carefully: what customer support is offered and how good is their customer service?

From a performance point-of-view, the two main factors will be the applications that the server will run and the number of users.

Applications have very different hardware requirements. In particular, email and database applications have high processing and storage requirements. Also consider that if a server is configured to perform multiple roles or support multiple software server instances (virtualisation), it will have to be that much more powerful, so that one application does not impact too much on the performance of another.

For users, the main tasks are to identify average and peak simultaneous access (the number of clients using an application at the same time) and storage requirements.

The main factors determining performance are the processor, system memory, speed of the disk subsystem, and network bandwidth.

Other specification decisions that may need to be made include:

■ Provision for fault tolerance (storage, power, and network connectivity).

■ Provision for data backup.

The timing of the installation project also needs to be considered with reference to wider business factors. Obviously the company needs to be doing well enough to afford the budget but you also need to be alert to other projects running in the business (product rollouts, recruitment drives, or training initiatives for instance) that could impact your project or which your new server project could affect (adversely).




Page 22

Integrating the Server

One of the first decisions to make about a new server is where to put it. Some of the issues to consider here are:

■ Environment - the server needs to be in a secure and climate-controlled environment. As a bare minimum, the server needs enough space and an adequate power supply.

■ Network access - the server needs to be as physically close as possible to the network that it serves. This could be a local network (intra-site) or WAN (inter-site) link. Network cabling imposes restrictions on how far apart equipment can be placed.

You should prepare plans indicating how the server will be located physically in the building and logically on the network. Diagramming applications such as Microsoft Visio can be used to create such plans. The next point is to work out how the server will integrate with the rest of the network. Interoperability will largely be determined by the operating system selected for the server. By far the simplest solution is to use a single operating system for all servers on the network, but this is not always possible. The two most important factors are:

■ Support for protocols - this means that the server is able to exchange data with other computers on the network.

■ Support for clients - this means that software is available for client workstations to use applications running on the server directly.

If machines are not directly interoperable, it may be possible to configure another machine to act as a gateway. However, this will add complexity to the network design and decrease performance.

Connecting a Windows client to a UNIX server using a client-based service

Connecting a Windows client to a UNIX server using a server gateway



Page 23

The server will also need to be configured for the network. The information required usually consists of:

■ Network protocol (including addressing and other configuration parameters)4.

■ Network name (domain suffix).

■ Computer (host) name.

As the computer name needs to be unique, there may be some convention in place for adding new computers. Some companies use character names from books or films but putting a structured, descriptive naming scheme in place can greatly assist configuration and troubleshooting5. A structured naming scheme might use fixed length codes identifying key attributes of the machine:

■ Physical Location.

■ Client or Server / OS.

■ Department.

■ Server Role or main user's Job Role.

■ Unique ID.

Ongoing Management

Most management issues centre on having personnel expert in the configuration of the operating system and applications software installed on the server, though some hardware systems may require specially trained staff. As well as providing staff that can administer the server, it also important to incorporate it into backup and other security procedures.

See Unit 4.2 for more information about routine maintenance.

4 More than one protocol can be installed if necessary but this can degrade performance and make the system less secure. Many Windows systems run NetBIOS over TCP/IP to support pre-Windows 2000 hosts (and some features of Windows networking where NetBIOS still performs some functions). 5 One drawback is that it can assist attempts to infiltrate (hack) the network as host names become quite easy to guess. The naming scheme could also give an attacker quite a lot of information about how the network is structured. It is important not to use the user name as the computer name.




Page 24

Pilot Program

A pilot program is a test program to ensure that the server functions as expected. In the case of a server installation, it would consist of making the new server available to a number of selected users (test group) who would then perform a number of representative tasks, during which the server is closely monitored to evaluate performance and identify faults.

Another type of testing involves using software to simulate heavy user access, allowing load testing of the server under extreme conditions.

Pilot programs are quite complex to set up, but may be considered necessary for any component or software upgrade, especially if the upgrade is taking place on a mission-critical system.

Server Upgrade Plans

Upgrading covers many different scenarios. In fact, you are unlikely to perform many completely new installations; most jobs involve replacing existing equipment. Some of the main upgrade activities are:

■ Replace an existing server with a better specified machine.

■ Replace or add components to a server.

■ Upgrade the software on a server to a new version or different vendor.

The need for an upgrade should be revealed well in advance by having an effective monitoring program in place. This program should also reveal precisely which components require upgrading. For example, if server performance seems sluggish, an uninformed reaction may be to upgrade the processor and add memory, whereas the problem could actually lie with inadequate network bandwidth.

See Unit 4.1 for more information about monitoring and optimising performance.

Upgrade jobs range from simple to complex, but for any job it is important to make an assessment of the risk factors involved in performing the upgrade.



Page 25

The main considerations when planning an upgrade are:

■ How to deal with the likely interruption of service.

■ Protecting data stored on the existing system and possibly migrating it to the new system.

■ Ensuring the upgraded components or software are compatible with the existing infrastructure.

Scheduling the Upgrade

An upgrade is likely to involve a certain amount of downtime (the period during which a server is inaccessible to users). Users must be informed when downtime is scheduled, how long it will last, and why the upgrade is being performed. If the upgrade project overruns, keep users notified on when they can expect services to be resumed.

You need to account properly for lost business time. If paying a technician overtime to work at the weekend seems expensive, compare it to the cost of having your sales department twiddling their thumbs for a few hours on Tuesday afternoon. In a global service-oriented organisation, there may be no out-of-hours periods, but you can use monitoring to determine when the server is least utilised. Mission-critical systems in organisations of this type are likely to have redundancy, allowing a server to be taken out of service without disrupting users.

As with new installations, upgrades should not be performed when other projects are running, if at all possible. For example, if a webcast with the company's main account is scheduled for Wednesday, choosing Tuesday to upgrade the media server isn't going to look like a good decision if anything subsequently goes wrong with the webcast. It is important to get clear lines of communication with and commitment from the business groups that will be affected by the change.

If there are several upgrade tasks to perform, it makes sense to roll them up into one job. An effective management and monitoring plan for the server will ensure that you can perform upgrades pro-actively and are not simply reacting to events.

Protecting and Migrating Data

It is vital that the configuration of the current system be well documented and understood by the installers. If you don't know how things work at the moment, the chances of unforeseen errors and configuration mistakes occurring are greatly multiplied.




Page 26

As part of the upgrade plan there should be a rollback plan. This will describe how the original configuration can be restored in the event that the upgrade cannot be made to work.

The best, but most expensive, method is to provide redundancy, so that other servers maintain the service while one is being upgraded.

There are various methods of providing redundancy, including cluster services and hot spares. See Unit 5.3 for more information.

If only one server is available, the first point of the rollback plan is to make a backup of the system configuration. The second, often omitted, step is to test the backup to ensure that it is valid and can be used to restore the system. The most convenient means of backup for this purpose is to use disk imaging software, such as Norton Ghost, Paragon Drive Backup, or Acronis True Image.

Implementing the Plan

Your approved project plan should show who is responsible for doing what, when they should be doing it, and for how much money. When working with a project team, it is essential to update progress on each task. Clear and open communications between the team members is vital, especially where one person's task affects the start or completion of another person's. Each installation task should be approached methodically and is likely to require the following actions:

1) On delivery, check that the correct components have been supplied and that the box contents are intact and complete, including cabling, connectors, screws, chassis keys, and other easy to lose items. Inspect components for any visible signs of damage (bent or broken pins on connectors for instance).

2) Read the necessary documentation and instructions and check the vendor's website for the latest information and software/driver versions.

3) Schedule downtime and notify users about the service outage.

4) Make and test a backup in line with your rollback plan.



Page 27

5) Prepare the work area, taking appropriate ESD (ElectroStatic Discharge) precautions. Perform the installation or upgrade, including installing any utilities or added software and service packs.

6) Check that the installation has succeeded and that the product is working.

7) If necessary, perform the migration of data from the old system to the new.

8) Test the installation.

9) Document the configuration and installation process. Make a note of any problems encountered (and resolved) - you should include any test plans and results and feedback from users and recommendations to future project teams.

Verifying the Plan

Following installations or upgrades, it is important to put the final steps of your installation plan into practice:

1) Test the system both for functionality and usability. Remember that the success of an upgrade will depend on users' ability to cope with the new software. Training and support needs to be provided to help users get the most out of new features. This will help to justify the budget allocated.

2) Assemble and file the service documentation (see below).

3) Establish the ongoing maintenance and support plan for the server, showing clear responsibilities and timescales. Some features of this will be:

Establish a monitoring program, including a baseline performance measurement.

Add the software to system backup routines; test that the backup can be used to restore data.

Identify any effects on related systems and procedures (for example, update a server's Emergency Repair Disk following any system change, consider whether adding an application will require more backup media, or update the disaster recovery plans to include the extra systems).




Page 28

Configuration Management

Configuration Management can be summed up as an answer to the question "If something changes, can we get back to the way it was?". In this respect, configuration management is related to change management. Change management itself can be summed up as an answer to the question "Is it a good idea to change this?"

ITIL Configuration Management Model

ITIL® (IT Infrastructure Library) is a popular documentation of good and best practice activities and processes for delivering IT services. Under ITIL, configuration management is implemented using the following elements:

Service Asset

Service assets are things, processes, or people that contribute to the delivery of an IT service.

Configuration Item

A Configuration Item (CI) is an asset that requires specific management procedures for it to be used to deliver the service. Each CI must be identified by some sort of label. CIs are defined by their attributes, which are stored in a Configuration Management Database (CMDB).

Baseline

The fundamental concept in configuration management is the baseline. The baseline represents "the way it was". A baseline can be a configuration baseline (the ACL applied to a firewall for instance) or a performance baseline (the throughput achieved by a particular server for example).

Configuration Management is not possible without baselines!

Configuration Management System (CMS)

The CMS is tools and databases that collect, store, manage, update, and present information about CIs.

One of the goals of the CMS is to understand the relationships between CIs. Another is to track changes to CI attributes (and therefore variance from the baseline) over time.



Page 29

Implementing Configuration Management

The main difficulty in implementing a workable configuration management system is in determining the level of detail that must be preserved. This is not only evident in capturing the asset database and configuration baseline in the first place but also in managing Moves, Adds, and Changes (MACs) within the server infrastructure.

In terms of server maintenance tasks, it is imperative that the ramifications of any change made to the server be considered in terms of the impact on the organisation as a whole. Therefore a CMS will require that configuration changes be made only when there is a valid job ticket authorising the change. This means that the activity of all server admins, whether it be installing new devices or troubleshooting, is recorded in job logs.

Consequently, configuration management involves drafting and enforcing policies and procedures (Standard Operating Procedures or SOPs) to govern change control at all levels of configuration and troubleshooting activity.

Regulations

Configuration management is important in terms of demonstrating due diligence.

Due diligence is a legal term meaning that responsible persons have not been negligent. Negligence may create criminal and civil liabilities. Many countries have enacted legislation that criminalises negligence in information management.

In the US for example, the passage of the Sarbanes-Oxley Act (SOX) has mandated the implementation of risk assessments, internal controls, and audit procedures. The act was introduced following a number of high-profile accounting scandals, including the collapse of Enron. As another example, the Computer Security Act (1987) requires federal agencies to develop security policies for computer systems that process confidential information. In the EU, data protection legislation requires companies to store and process personal data securely.

A civil liability means that in the case of a security incident where liability cannot be assigned to the perpetrator or damages cannot be recovered, those responsible for security (such as directors and senior managers) may be forced to pay damages to third parties (such as customers or suppliers affected by the incident).

A Configuration Management System demonstrates that the network is being managed competently and securely.




Page 30

Service Level Agreements

A Service Level Agreement (SLA) is a promise to provide a specific service to a specific standard which has been agreed between two organisations.

In terms of maintaining and troubleshooting servers, your organisation is bound to fulfil the terms of the SLA. Maintenance activity or troubleshooting is highly likely to impact on the SLA and therefore the advice of stakeholders within your company and within client companies may need to be sought.

Most SLA's will allow for a defined period of unexpected downtime and also for a defined period of offline maintenance. Given that, it is still vital to communicate the occurrence and duration of known service outages and to acknowledge unexpected downtime. It is also vital to communicate when services have been fully resumed.

Key Performance Indicators

Key Performance Indicators (KPI) are metrics used in quality and performance management and are also likely to be quoted in an SLA. Some of the main KPIs relative to service availability are as follows...

Availability / Uptime

This is the percentage of time that the system is available and working, measured over the defined period (typically one year). The corollary of availability is downtime (that is, the percentage or amount of time during which the system in unavailable).

Full availability is usually loosely described as 24x7 (24 hours per day, 7 days per week) or 24x365 (24 hours per day, 365 days per year). For a critical system, availability will be described as "two-nines" (99%) up to five- or six-nines (99.9999%):

Availability Annual Downtime

99.9999% 00:00:32

99.999% 00:05:15

99.99% 00:52:34

99.9% 08:45:36

99.0% 87:36:00

Downtime is calculated from the sum of scheduled service intervals (Agreed Service Time) plus unplanned outages over the period6.

6 Downtime may or may not be defined to include agreed service time. It is important to verify this point.



Page 31

Response Time

This is a measure of the mean time taken to acknowledge a support request (but not necessarily to fix the problem).

Mean Time Between Failure (MTBF)

This represents the expected lifetime of a product or system. If less than the defined period (of the SLA), you can work out the number of expected outages that will be experienced.

Mean Time to Repair (MTTR)

This is a measure of the time taken to correct a fault so that the system is restored to full operation.

Mean Time Between Service Incidents (MTBSI)

This represents the time between the point of failure of a system and its next point of failure (that is, MTBF + MTTR).

Documentation

The purpose of documentation in terms of configuration management is as follows:

■ Identify each component (CI) and label it.

■ Capture each CI and its (relevant) attributes in a CMDB.

■ Capture relationships between CIs - this is best done using diagrams.

■ Capture changes to a CI as a job log and update the CMDB.

Identification and Labelling

The more servers you have in an environment, the more important an effective labelling system becomes. Labelling is a difficult thing to get right:

■ The label has to be long enough to identify the server but short enough to read easily.

■ The label convention has to be flexible enough to cope with growth and new server roles (while you might have to re-label individual servers now-and-again, you don't want to have to re-label everything).




Page 32

The simplest system is to number each rack and then each server in each rack. You may want to try to capture additional information, such as host name, OS, IP address or subnet, or distinguishing servers hosting virtual machines from standalone servers. You might be able to use colour-coding to organise servers into categories.

Make sure you label the front and the back of the server. Put the label in the same place on each server. It is also a good idea to add the label to the server's desktop wallpaper so that it is clear which server you are operating once you have logged on (using a KVM switch or remote management tool).

As well as the servers, it is also important to label network and power cabling so that if you have to pull a plug, you pull the correct one.

The ANSI/TIA/EIA 606 Administration Standard for Commercial Telecommunications Infrastructure sets out a basic scheme for documenting a typical office network installation7. It includes suggested identifiers for different subsystems and elements of the installation, colour-coding of cables, distribution frames, and ports designed for different applications, and symbols for representing the different parts of the infrastructure in diagrams and schematics.

Whatever the precise system used, each bit of the network infrastructure should be tagged with a label ID.

In addition to the label ID, a server will also have a vendor service tag, used to obtain warranty support and download patches and drivers from the vendor's support site. There may also be a separate asset tag used to identify the server in your company's accounting database.

Schematics and Diagrams

Diagrams are the best way to capture the complex relationships between the hardware sever infrastructure and the server software roles (plus virtual machines, if the infrastructure is virtualised). Diagrams can be used to model physical and logical relationships at different levels of scale and detail.

A schematic is a simplified representation of a system. In terms of the physical network topology, it can show the general placement of equipment and telecommunications rooms plus device and port IDs without trying to capture the exact position or relative size of any one element.

7 The standard has recently been updated with an addendum (T606-A-1) covering rack and server labelling in a data centre.



Page 33

Schematics can also be used to represent the logical structure of a network and placement of various servers, in terms of zones and subnets. In this sense, schematics can use either representative icons or pictures or drawings of actual product models.

Schematics can either be drawn manually using a tool such as Microsoft Visio or compiled automatically from network mapping software.

Ipswitch WhatsUpGold network mapping software (www.whatsupgold.com)

Rack Diagram

A rack diagram shows the location and ID of racks in the server room or data centre plus the ID of each piece of equipment installed in the rack.

Network Mappers

Network mapping means gathering information about the way the network is built and configured and the current status of hosts. The following information is all of use:

■ Protocols, services, and applications running on the network.

■ Host workstation and server OS types and patch status.

■ Network addresses and host names.

■ Network interconnect device types and status.

■ Network security appliances and software.

■ User accounts and groups (especially administrative / root accounts) and passwords.

This sort of information may be gathered by network management software, such as CiscoWorks, Microsoft’s System Center, HP’s OpenView, or numerous other third-party products.




Page 34

Service Documentation

The following service documentation should be created and maintained for each server commissioned:

■ Installation log - installer and date of installation, vendor specification and warranty details (plus location of documentation, technical support contacts, service tag, and so on), NOS and applications software, intended server role, and responsible administrator(s).

■ Configuration log - RAID configuration, custom configuration for any applications, details of services available (plus startup type - manual or automatic), network configuration, and performance baseline.

■ Service log - details of update testing and patches applied8, any hardware or software upgrades or repairs performed, updated performance baselines, plus additional useful troubleshooting information.

■ Decommissioning log - verify software licence removal and data disposal or archiving.

You need to keep the same sort of logs for each virtual server too.

If the server gains a new role, you should add that information to the relevant log. If the server is repurposed in a completely different role, it is probably best to start a new set of logs.

Vendor-specific Documentation

As well as installation and service logs, you should keep vendor documentation available to assist with maintenance and troubleshooting. Always familiarise yourself with the user or service guide before starting on hardware maintenance or upgrade. The vendor will have listed supported upgrades and configurations for things like memory and RAID. You should also be able to find "best practice" information about maintenance and troubleshooting. Many vendors also supply troubleshooting diagnostic tools or checklists specific to a model of server.

Use the vendor's support website to obtain copies of user guides plus drivers and troubleshooting information. Make sure invoices and warranty support information is available so that technicians know the appropriate support channels available (phone, email, text chat for instance) and the terms of warranty work (onsite or back-to-base for instance).

8 If a patch is released and not applied because it causes problems with an application or hardware, it is very important to record why the patch has not been applied and what security or performance issues this could raise.



Page 35

Implementing a CMS, CMDB, and Job Logs

The implementation of the administration system could be paper-based but computer records are much easier to search and update. The following tools could be used for a configuration management system:

■ Spreadsheet - something like Microsoft Excel can quite comfortably store baseline CIs on a small network though it is less suitable for tracking jobs and changes to the configuration.

■ Custom database - there are a number of dedicated applications for implementing CMS. While they are correspondingly expensive and difficult to configure, these provide the best means of preserving accurate information and controlling procedures through job ticketing.

CiscoView Device Manager, part of the CiscoWorks suite

■ Diagrams - Microsoft Visio is widely used to prepare network schematics.

As with any communications, job tickets should be completed professionally, with due regard for spelling, grammar, and clarity. Remember that other people may need to take action using just the information in the ticket and that analysis of tickets will take place as part of quality assurance procedures. It is also possible that tickets will be forwarded to customers as a record of the jobs performed.




Page 36

Secure Documentation

The information contained in the configuration management documentation would be extremely useful to anyone seeking to attack the network. Any details of system configuration or errors that have been experienced would help an attacker to penetrate the security systems set up to defend the network or individual servers.

Consequently, the configuration management documentation should be secured against snooping. Printed documents could be stored in a secure room or safe while electronic documents should be encrypted and protected by file access control permissions.

See Unit 2.5 for more information about security systems.

Another point to be made is about storage of passwords for administrator-level accounts. Generally speaking, administrative privileges should be delegated to personal accounts and the account holder be made responsible for the security of the account (that is, making sure that only the account holder knows his or her password). This means, among other things, that the password should never be written down or stored in any sort of computer file.

Some network devices, support websites, or web hosting packages do not support personal accounts however so it is often necessary to use shared administrator accounts and to make the same password known to several different technicians. Shared admin passwords should be avoided where possible but if they are absolutely necessary, it is best to store them using password manager software. The password manager encrypts the file and restricts access to authorised users. Each technician would need to access the password manager using their personal authentication credentials and the security system should log access attempts.

Passwords should not be recorded in routine systems documentation.

Do not reuse admin passwords for different sites. For example, if you have a support account with a server vendor, you should never use a password on that account that is also used on your own network. It might seem tempting to reduce the number of admin passwords that have to be remembered, but repeating a password somewhere else means that your security will then only be as good as the remote site's security.

Don't forget that if a technician leaves the company, you need to change any shared passwords that the technician may have known.



Page 37

Equipment Disposal

Equipment disposal or decommissioning requires its own process. The principal issue is that of remnant removal. Remnant removal refers to fully erasing decommissioned media (including old hard disks, removable disks, flash drives, tape media, CD and DVD ROMs, and so on). The problem has become particularly prominent as organisations recycle their old servers and PCs, either by donating them to charities or by sending them to a recycling company, who may recover and sell on parts. There are at least three reasons that remnant removal is critical:

■ An organisation's own confidential data could be compromised.

■ Third-party data that the organisation processes could be compromised, leaving it liable under Data Protection legislation (in addition to any contracts or Service Level Agreements signed).

■ Software licensing could be compromised.

The main issue is understanding the degree to which data on different media types may be recoverable. Data "deleted" from a magnetic-type disk (such as a hard disk) is not erased. With the proper tools, up to 60% of a drive's deleted information can be recovered. There are several approaches to the problem of data remnants on magnetic disks:

■ Full format using the OS-supplied utility - this is the simplest method but is not completely reliable.

■ Overwriting - disk wiping software ensures that old data is destroyed by writing to each location on the media in a random pattern, usually multiple times. This is suitable for all but the most confidential data, but is time consuming and requires special software.

■ Destruction - a magnetic disk can be mechanically shredded or degaussed or even obliterated using acid. Obviously, this is costly and renders the disk unusable (so it cannot be recycled).

■ Use a reputable recycling company - any of the above methods could be contracted out to a third-party, but this requires a degree of trust in the third-party to erase the data properly.

■ Disk encryption - this method encrypts all the information on the disk, so that any remnants could not be read. This requires third-party software under current versions of Windows and there will be an impact on performance.

Optical media cannot be reformatted. Discs should be destroyed before discarding them. Shredders are available for destroying CD and DVD media.




Page 38

Review Questions Module 1 / Unit 2 / Configuration Management

Answer the following questions. The correct answers are in the accompanying "Labs and References" manual.

1) What is a SMART objective? An objective that is well-drafted: Specific, Measurable, Achievable, Realistic, Timely.

2) What is a project stakeholder? Those who can impact a project or could be impacted by it and can influence its outcome.

3) What three main stages would you consider when planning for a new server installation? Specify server hardware and software, identify how it will be integrated into the network, and make a management plan.

4) What is a pilot program? A test program to ensure that a new server (or server upgrade) can be made operational without risking damage or disruption to network services or data.

5) What three things should you consider when planning a server upgrade? Compatibility (whether the new hardware or software is compatible with the existing hardware and software), scheduling (and informing users), and protecting data (having a rollback plan).

6) Why would a MAC require a job ticket? In the context of configuration management, a Move, Add, or Change (MAC) should be properly authorised and documented.

7) How might SLAs affect server upgrade plans? A Service Level Agreement is a commitment to provide a certain level of service to customers. The upgrade work needs to be planned carefully to ensure it does not breach SLAs.

8) True or false? The local administrator or root user password should be recorded in the system installation log to ensure the password is never lost, which would permanently deny access to the system. False - passwords should not generally be written down. If they must be shared it should be through a secure password management system.


Study Notes Storage and RAID

Page 39

Module 1 / Unit 3 Storage and RAID

Objectives


■ Describe the features and performance characteristics of server-class hard disks and solid state drives.

■ Describe features of the SCSI interface and know how to configure devices correctly.

■ Describe features of the SATA and SAS interfaces and know how to configure devices correctly.

■ Select an appropriate RAID level for a given storage solution.

■ Configure a RAID array.

Hard Drives

Like a desktop PC, a server will be configured with one or more hard drives to store the operating system and applications software. Disk space may also be provided on a file server as a shared resource for network users.

Unlike most desktop PCs however, server disks will typically be faster for higher performance and add resilience features such as RAID - the ability for multiple hard disks to work together to increase access speed to files or provide redundancy in case one of the drives fails.

All new servers use disk interfaces based on either Serial Attached SCSI (SAS) or Serial Advanced Technology Attachment (SATA). Some older servers may be configured with parallel SCSI interfaces.

Drives are available in two sizes: 3.5" Large Form Factor (LFF) or 2.5" Small Form Factor (SFF). SFF now dominates the market, with LFF drives gradually being phased out.




Page 40

HDDs and SSDs

Most hard disks are of the mechanical type (HDD), where data is stored magnetically on coated metal or glass platters with drive heads to read and write the data. The drive heads must be moved to the required location on the disk, which introduces a degree of latency over and above the time taken for the controller to transfer data, particularly when the data is fragmented across different disk sectors. The performance of a magnetic hard drive is mostly determined by the speed at which the platters are spun (7.2K, 10K, or 15K rpm [revolutions per minute])9.

Recently, flash memory based drives (Solid State Drives [SSD]) have been introduced to the market10. As these have no moving parts, they generally use less power and are lower latency, especially when reading non-sequential data. However, write times can be slower for SSDs than for HDDs. Also, the NAND memory cells used to store data only support a limited number of write/erase cycles, which means that the drive will become less reliable as it ages. To compensate, the OS and the drive controller firmware should perform wear levelling, to distribute writes over different cells.

HP SSD with SATA interface

The advantages and disadvantages of SSDs mean that they are often deployed in servers where there are low disk write requirements. For example, SSDs are better suited to front-end web servers and HDDs to file servers.

9 A 15K drive should support an internal transfer rate of up to about 180 MBps while 7.2K drives will be around 110 MBps. 10 There are broadly two types of SSD: Single Layer Cell (SLC) and Multi Layer Cell (MLC). SLC stores 1 bit per cell while MLC stores 4 or more bits per cell, yielding higher capacities at lower cost. MLC can be slower however as it requires substantial error correction processing. It is also perceived as not reliable enough for enterprise server applications as it supports fewer write cycles, though improvements in the technology may change that perception over time.



Page 41

SSDs also carry a significant price premium over HDDs and rise almost exponentially in cost at sizes over 256 GB.

SSDs use the same interfaces as traditional hard drives (that is, SAS or SATA).

Enclosures and Backplanes

A drive is housed in an enclosure or caddy. The drive may then be connected to the controller via a cable but on enterprise-class servers it is more usual for the enclosures to be connected via a backplane. Rather than using cabled connectors, the drives plug (or "mate") into a combined data and power connector on the enclosure. This means that drives can be easily added and removed from the front of the case without having to open the chassis.

Enclosure and backplane on an HP server - 1) The drive mates with the port on the backplane card; 2) Data and power cables on the other side of the backplane card

connect to the drive controller and PSU

The drives are secured and released from the server using a latch. Many server drives are hot-swappable, meaning that they can be added or removed without powering down the server.

1 2

2




Page 42

The SCSI Interface

The Small Computer System Interface (SCSI) has been in use as an expansion bus interface since the 1980s. There have been many revisions to the standard. Originally a parallel interface, SCSI is now used for serial connections for disk drives (Serial Attached SCSI) and peripheral devices (Firewire or IEEE1394).

In a PC server, a SCSI adapter (known as the Host Adapter) may be integrated with the motherboard or installed as a PCI-X or PCIe expansion card.

SCSI supports multitasking and multithreading through disconnect-reconnect. When a device experiences a delay in processing a request, it can release control of the bus (disconnect), allowing other devices to use it. When the device is ready to transfer data, it reconnects.

More information about SCSI can be located at the T10 Committee website (www.t10.org) and the SCSI Trade Association (www.scsita.org).

Parallel SCSI Standards

SCSI-1

SCSI-1 is sometimes simply known as SCSI; if a number is not mentioned you can usually assume that SCSI-1 is being referred to. SCSI-1 is defined as an 8-bit bus with a 5 MBps transfer rate. Up to 7 SCSI devices can be daisy-chained to a single SCSI port.

SCSI-1 devices can be used with devices that use the higher transfer rates of the more advanced SCSI-2 protocols, but will cause the whole bus to slow down. Practically speaking therefore, SCSI-1 devices are obsolete.

SCSI-2

The SCSI-2 specification was approved by ANSI in 1990. It is an extensive enhancement of the original standard and defines support for many of the more advanced SCSI features still in use today.

SCSI-2 includes the following enhancements to the original SCSI-1 specification:



Page 43

Type Notes

Fast SCSI A high-speed transfer protocol doubles the speed of the bus to 10 MHz, resulting in a transfer rate of 10 MBps with 8-bit regular SCSI cabling or even higher when used with Wide SCSI (referred to as Fast-Wide SCSI).

Wide SCSI Widening the original 8-bit SCSI bus to 16 bits permits more data throughput at a given signalling speed.

More devices per bus

A bus that runs with Wide SCSI supports 16 devices (including the host adapter).

Improved cables and connectors

SCSI uses a confusingly large number of different cable and connectors. SCSI-2 defines new higher-density connections.

Active termination

Termination is an important technical consideration in setting up a SCSI bus. SCSI-2 defined the use of active termination, which provides more reliable termination of the bus.

Command queuing

One of SCSI's strengths is its ability to allow simultaneous multiple outstanding requests between devices on the bus.

SCSI-3

SCSI-3 is not so much a formal standard as a number of new technologies and sub-standards introduced under the SCSI banner. Some features conflict with each other or represent different approaches to how SCSI is to be implemented or used.

One of the key changes is the use of serial signalling in some products, causing previous versions of SCSI to be retroactively labelled SCSI Parallel Interface (SPI) or "parallel SCSI".

Product Notes

Improved cabling

Improved cabling for the use of Wide SCSI. The use of HVD or LVD signalling and termination allows for greater cable lengths.

Ultra SCSI Doubling the SCSI-2 system bus speed to 20 MHz, meaning 20 MBps with 8-bit SCSI and more with Wide SCSI.

Ultra2 SCSI Another doubling of the system bus speed to 40 MHz. No support for SE signalling.

Ultra3 SCSI Double Data Rate (DDR) signalling, effectively doubling transfer rates, plus improvements to bus communication and arbitration procedures.

Firewire A protocol standard for one type of Serial SCSI used for peripheral device expansion and consumer electronics.

Serial Attached SCSI (SAS)

Serial, point-to-point interface (similar to PCIe) supporting the SCSI command set.

Fiber Channel (FC-AL)

Specification for a Storage Area Network (SAN) architecture supporting the SCSI command set.

iSCSI Use of the SCSI command set over IP-based networks.




Page 44

The various "flavours" of Serial SCSI support hot-swapping of devices; some types of parallel SCSI also support hot-swapping.

Summary of SCSI Types

Interface Devices

(Excluding

Host

Adapter)

Rate Max Cable

Length (m)

Connector

SE LVD HVD

SCSI-1 7 5 MBps 6 12 25 50-pin

Fast SCSI

7 10 MBps 3 12 25 50-pin

Fast-Wide SCSI

15 20 MBps 3 12 25 68-pin

Ultra SCSI

7 20 MBps 1.5 12 25 50-pin

Wide Ultra SCSI

15 40 MBps - 12 25 68-pin

Ultra2 SCSI

7 40 MBps - 12 25 50-pin

Wide Ultra2 SCSI

15 80 MBps - 12 25 68-pin / 80-pin

Ultra3 SCSI (Ultra160 SCSI)

15 160 MBps - 12 - 68-pin / 80-pin

Ultra320 SCSI

15 320 MBps - 12 - 68-pin / 80-pin

Ultra640 SCSI

15 640 MBps - 12 - 68-pin / 80-pin

Firewire Serial (63 devices)

100 - 800 Mbps

- 4.5 - 6-pin / 9-pin

SAS Serial (128+ devices)

3 - 6 Gbps - 6 - Various

FC-AL Serial (127+ devices)

1 - 4 Gbps - 500+ - Various

The number of devices listed in the table excludes the host adapter. So for example, narrow SCSI supports 8 devices including the host adapter and wide SCSI supports 16. Note that Firewire, SAS, and Fibre Channel are serial links and so measured in megabits or gigabits per second rather than megabytes.



Page 45

Installing the Host Adapter

The majority of servers will incorporate a SCSI host adapter (most likely to support Serial Attached SCSI) on the motherboard. If SCSI is not supported on the motherboard (or if an upgraded SCSI / RAID solution is required), a host adapter or storage controller must be installed in a suitable expansion slot (typically a x4 or x8 PCIe slot or PCI-X slot).

Remember that SCSI throughput is restricted to the throughput of the local bus. For example, a host adapter connected to a version 1 PCIe x1 slot would only have 250 MBps bandwidth and would not be able to take advantage of Ultra 320 SCSI speeds.

Connecting the SCSI Devices

Parallel SCSI devices may be internal, external or a combination of both. Many configurations are possible. You can connect disk drives, tape devices, CD-ROM drives, and scanners.

Internal devices are connected to a single piece of flat ribbon cable and require a connection to the power supply or (in the case of most drives) plug into a backplane.

External devices are daisy-chained. A cable is connected to the external port at the back of the host adapter, then to one of the two ports at the back of the first external device. If a second device is required, another cable is connected to the first device, then to one of the two ports on the second device and so on. External devices normally require a connection to the mains supply and should be powered up before you turn on the server.

Internal devices - ribbon cable




Page 46

External devices - daisy chained

Internal and external devices

SCSI Signalling

In addition to the different parallel SCSI standards, there are three types of parallel SCSI signalling:

■ Single-Ended (SE - the original specification).

■ High Voltage Differential (HVD).

■ Low Voltage Differential (LVD).

Each type places signals on the cabling differently.

SCSI systems use a parallel data path and can suffer from data skew (when the signal on each wire travels at different speeds). This limits cable lengths to 3m (10 feet) for fast SCSI or 6m (20 feet) for standard SCSI using SE signalling.



Page 47

Differential SCSI uses two signal paths for each data line and the method by which data is transmitted makes it much less susceptible to corruption. Differential signalling can either be high-power (to achieve longer distances) or low power. In the event, LVD proved much more popular, to the extent that HVD is very rarely found on devices designed for the PC server marketplace.

SE and LVD devices can be combined on the same bus, but the SE maximum bus length then applies and performance of LVD devices can be reduced, as the whole chain operates at the speed of the SE device (unless you use a bus expander connected to the host adapter to separate SE and LVD devices).

HVD devices must not be placed on an SE or LVD bus.

Symbols for SE, HVD, LVD, and LVD/SE (multi-mode) SCSI ports

SCSI Connectors and Cabling

Internal SCSI Connectors

Older internal SCSI devices are generally connected by 50 or 68-way ribbon cable depending on whether the bus is narrow or wide11. The connector on the device is either a male 50-pin IDC (Insulation Displacement Connector)12 or a female 68-pin High Density (HD) connector.

Male and female IDC50 connectors

Male HD68 connector

Most SCSI hard drives now use the Single Connector Attachment (SCA)13. This 80-pin adapter combines the functions of data and power connector and enables autoconfiguration of SCSI settings, such as the ID and termination, and hot-swapping, plus "sideband" signals, such as LED and status monitoring. Such drives slot into a backplane rather than use cabling. The connector on the drive is male.

11 These connectors were also used with backplanes to implement drive arrays. 12 It is possible to obtain 68-wire cable to 50-pin connector adapters. 13 There are two versions: SCA-1 and SCA-2. SCA-1 is obsolete.




Page 48

External SCSI Connectors

External SCSI devices can be connected to the PC-based controller through various types of connector:

■ SCSI-1 - DB25 or (more commonly) 50-way Centronics "edge" connectors.

Male and female DB25 connectors

Male and female 50-pin Centronics connectors

■ SCSI-2 and -3 - High Density DB connectors (50-pin and 68-pin) or Very High Density (Micro-Centronics) 68-pin connectors.

High Density 68-pin male and female connectors

Very High Density 68-pin male and female connectors

Centronics connectors use clips to secure the connector; other types are screw-in. The cabling is round and often quite thick because of the heavy shielding used.

Terminating the SCSI Devices

A parallel SCSI bus must be terminated correctly in order to work. Termination absorbs the signal passing along the bus so that it does not reflect back onto the bus and corrupt valid signals.

You must install manual terminators or enable termination on the devices at each end of the SCSI bus. Only SCSI-1 devices require manual termination; the later specifications are self-terminating (automatic termination).

All terminators must be powered. Power can be supplied from the device itself, from the host adapter, or from the backplane (if present).



Page 49

Types of Terminator

Different iterations of SCSI have used different termination methods. If all the devices on the bus are the same type, configuration is usually straightforward, but when there is a mix of devices, things can be more complex.

■ Passive terminators - these are simple devices used with SCSI-1 and SE signalling14. A terminator pack may be internal to the device and enabled or disabled by a jumper or switch or an external resistor pack may need to be fitted.

■ Active terminators - these contain circuitry to regulate the voltage of the termination current, making them more reliable devices. Active terminators are recommended for SE signalling and required for differential signalling. Active terminators are always built into the device and in many cases will be configured automatically by the SCSI driver software.

■ Active negation terminators - these contain circuits to cope with fast bus speeds and are required for Ultra SCSI and faster.

■ Multimode - normally a terminator must match the type of signalling in use, but a multimode (or LVD/SE) terminator is one that can be used with both SE and LVD signalling15. An LED should indicate what type of signalling is in use.

■ High-byte - if a wide bus contains narrow devices (or a wide device is connected to a narrow bus), the "wide" or "high-byte" signals must be terminated at the point where the bus changes from wide to narrow (typically this will be implemented in the cable adapter).

Enabling and Configuring Terminators

Termination may involve adding a special terminator plug to the open port on the back of the last external device, implementing resistor packs on the device, setting jumpers or switches, crimping a terminator adapter to cabling, or running a setup program. Check your documentation for each device to see how it is terminated.

Host Adapter Termination

The host adapter should be terminated if there are only internal or only external devices attached to it. If both internal and external devices are attached to the host adapter, you should disable termination on the adapter itself and enable termination on the ends of the internal and external chains.

14 A passive terminator is essentially just a resistor; it absorbs the signal. This method is not always completely reliable, especially over longer cable runs. 15 It may also simply be described as providing "LVD Termination"; in which case confirm that it is multimode. You may also see references to "Forced Perfect" terminators; these are used with SE devices to provide better performance than an active terminator.




Page 50

Correct termination of a mix of internal and external SCSI device chains

A guide to SCSI configuration can be found at: www.scsita.org/aboutscsi/SCSI_Termination_Tutorial.html

Setting SCSI IDs

A SCSI ID is an identifying label (a number from 0 to 7 [or 0 to 15 for wide SCSI]) assigned to each device. Each device on a SCSI bus must have a unique ID, including the host adapter, which is usually set to 7 or 15. Many older host adapters also insist that the bootable device (usually a hard drive) be set to ID 0.

SCSI IDs are necessary because each device must be uniquely identified when using the SCSI bus or when another device is sending a command. The SCSI ID determines which device has priority. When two or more devices are trying to use the SCSI bus at the same time, the highest ID is given priority (hence the use of 7 for the host adapter).



Page 51

On a wide bus, 7 is still the highest priority ID (it runs 7 to 0 then 15 to 8).

Setting IDs usually involves either physical or software setup. Many external devices have a simple mechanism on the outside for setting the ID, such as a dial or a window displaying the number, and buttons to change it.

Hard drives tend to have their IDs set using jumpers. These jumpers are commonly marked as A0, A1, A2, and A3 on the drive circuit board. The jumpers represent the binary increments 1, 2, 4, and 8 respectively. By combining the jumpers, any ID can be set between 0 and 15.

For example:

Jumper A3 A2 A1 A0

Represents 8 4 2 1

Setting OFF ON OFF ON

Binary digit 0 1 0 1 =5 (SCSI ID5)

Setting OFF OFF ON OFF

Binary digit 0 0 1 0 =2 (SCSI ID2)

Plug-and-Play SCSI or SCSI Configured AutoMatically (SCAM) automates the assignment of device IDs and allows the SCSI bus to dynamically shift and reallocate IDs when a new device is added to it. It also allows for the automation (or simplification) of termination.

Logical Unit Number (LUN)

Some parallel SCSI devices can perform more than one function (an auto-loading tape drive for instance). In this case, each function must be allocated a Logical Unit Number (LUN) from 0 to 7 or 0 to 15. This is normally assigned by the manufacturer.

Longer (64-bit) LUNs are also used in Storage Area Networks (SAN) based on Fibre Channel or iSCSI. See Unit 2.3 for more information about SANs.




Page 52

SCSI Drivers and Configuration Utilities

When installing a SCSI drive array onto a new computer, the adapter vendor will provide one or more utilities to configure the array. For example, an Adaptec host adapter is provided with the following configuration tools:

■ Array Configuration Utility / SCSI Select - press CTRL+A when prompted during the boot sequence to use this graphical menu-based program (stored in the adapter's firmware).

■ Adaptec Storage Manager - boot from the supplied CD / DVD to run this graphical utility.

■ ARCCONF - a command-line utility that can be used to create batch configuration files.

You can use these tools to configure SCSI bus settings (such as host adapter ID, termination, enable disconnect/reconnect, and so on) and configure a RAID array level (see later in this unit).

With the host adapter installed and the correct SCSI configuration applied, device driver installation for non-hard disk devices can be configured through the operating system or vendor setup utility as normal.

Serial Attached SCSI (SAS)

Serial Attached SCSI is the next generation of SCSI interface. It uses a serial interface with full-duplex communication over 2-pair wiring (much like PCI Express) but retains support for the SCSI command set.

SAS components can be rated at 3 Gbps or 6 Gbps.

A significant feature of SAS is support for thousands of devices (up to 16,384), using an Ethernet switch-like device called an expander. Each device is identified by a unique, manufacturer-coded ID, so there is no manual configuration to be performed. Also, SAS does not require termination, removing another complex configuration issue.

SAS also goes some way to uniting the SATA and SCSI standards. It provides both hardware support (the same connectors and cable) and software support (through the SATA Tunnelling Protocol) for SATA drives. This offers the opportunity to mix low-cost SATA drives with high-cost, high-performance SAS drives in an integrated storage solution, providing more flexibility than current models.



Page 53

Serial Attached SCSI Connectors

SAS devices feature a combined 7-pin data port16 and 15-pin power port (though some drives may also have a legacy 4-pin Molex power port too).

SAS connections are typically either single-lane (a simple adapter to device connection) or multi-lane (a single port on the adapter is connected to four devices).

Ports can either be "straight-through" or flush / surface mounted with right-angle connectors. Flush-mounted fittings are used in blade servers and other systems with restricted space. SAS backplane connectors are designed to "blind-mate", which means that a connection is made reliably when a drive caddy is inserted into a backplane. The design of the pins also reduces the chance of damage through ESD or a power spike (the full power pins connect after the other pins).

Hot-pluggable drives are not screwed into the chassis but slot into a drive cage. The cables for the devices connect to the drive cage (backplane) rather than the drive units.

Hot-pluggable drives on HP Proliant server

When removing a drive, you will probably need to use a utility or the OS to stop the device. This completes any cached write operations and prevents the NOS from trying to write data to the device while it is being removed. You also need to power down the device, either using software or a switch on the drive bay. Indicators on the drive bay should show when a drive is safe to remove. The drives are physically released and inserted using a lever or latch mechanism.

Unused drive slots should be filled with blanks to maintain the correct airflow and cooling within the chassis.

A number of connectors are associated with the use of SAS host adapters and drives:

16 In fact, on most SAS devices there are two data connectors (for redundancy). This is referred to as "dual-port". Obviously the backplane or cable must also support a redundant connection for this to work.




Page 54

■ SFF 8087 - internal mini connector for both adapter card ("host") and 4 drive or backplane ports ("target").

■ SFF 8088 - external mini connector for both adapter card and drive enclosures.

8087 and 8088 are newer connectors (with the product name Molex iPass) and will support increased SAS bus speeds when standardised (rated up to 10G).

■ SFF 8484 ("Multilane") - legacy 32-pin internal or external HD connector (host) supporting four 7-pin lanes (target). The remaining pins are used for "sideband" signals (LEDs and status monitoring).

■ SFF 8470 ("Infiniband") - legacy jackscrew connector for both internal and external use.

■ SFF 8482 - internal connector compatible with both SAS and SATA drives. This type of connector would be used principally to attach SATA devices (such as DVD drives) to an SAS bus.

The Serial ATA Interface

Serial ATA (SATA) was developed to address the limitations of the now obsolete parallel ATA or IDE interface. SATA would be used on low-end server hardware as a cheaper option than SAS.

4 SATA motherboard ports in front of an IDE port on an Intel motherboard

As the name suggests, SATA transfers data in serial format. This allows for thinner, longer, more flexible cables (up to 1m [39"]) with smaller, 7-pin data connectors. Each port supports a single device.

SATA cable for HP workstations



Page 55

The original SATA standard (SATA 1.5 Gbit/s or SATA/150) supports speeds of up to 150 MBps. This standard was quickly augmented by SATA 2.0, which specifies a 3 Gbps (300 MBps) transfer rate, and then SATA 3.0, running at 6 Gbps.

Other additions in SATA 2.0 include the use of port multipliers, which allow up to 15 drives to be connected to a single SATA adapter, and Native Command Queuing (NCQ), which enables the drive to analyse read/write operations and perform them in the most efficient manner, depending on the location of data on the disk

SATA 6 Gbps adds some extensions to NCQ to support isochronous data transfer (prioritising real time data such as video to ensure smooth playback).

There is also an eSATA standard for the attachment of external drives, with a 2m (78") cable. The main drawback of eSATA compared to USB or Firewire external drives is that power is not supplied over the cable. This is not so much of an issue for 3.5" drives, which require a separate power supply anyway, but limits the usefulness of eSATA for 2.5" portable drives. This drawback is addressed by the eSATAp standard, which uses a different port and connector to eSATA.

More information on SATA standards can be obtained from www.sata-io.org.

Hot Swapping

One of the major advantages of SATA over PATA is the support for hot swapping and consequently better compatibility with RAID configurations. Serial ATA 15-pin power connectors have been redesigned to provide support for both hot plugging and a 3.3V power supply in addition to the usual 5V and 12V.

Many drives retain a 4-pin Molex port for compatibility with legacy power supplies). Molex-SATA conversion adapters are also available.

SATA power connector

SATA and SAS

As mentioned earlier, SAS includes hardware and software support for SATA devices. The reverse is not true however; SAS devices cannot be plugged into an SATA bus.

A single enclosure could contain a mix of SAS and SATA devices, though a mix of SATA and SAS could not be used in the same logical volume. Mixing 3 Gbps and 6 Gbps devices is generally not recommended as it may impair 6 Gbps operation.




Page 56

Drive Arrays (RAID)

With RAID (Redundant Array of Independent Disks), many disks can act as backups for each other to increase reliability and fault tolerance or they can act together as one very large drive.

RAID can also be said to stand for "Redundant Array of Inexpensive Disks" and the "D" can also stand for "devices".

The RAID advisory board defines RAID levels. The most common levels are numbered from 0 to 6 where each level corresponds to a specific type of fault tolerance. There are also proprietary RAID solutions.

RAID Level Fault Tolerance

Level 0 Striping without parity (no fault tolerance)

Level 1 Mirroring/duplexing

Level 2 Striping with ECC (Error Correction Code)

Level 3 Striping with a dedicated parity disk

Level 4 Independent data disks with shared parity disk

Level 5 Independent data disks with distributed parity blocks (striping with parity)

Level 6 Second parity

In addition to these primary levels, it is also possible to "nest" one RAID solution within another (for example, you could mirror two stripe sets to boost performance without sacrificing fault tolerance).

RAID Performance

To choose an optimum RAID solution, you need to balance the server role with factors such as cost, performance, and security. Higher RAID levels usually require more costly controllers but may be able to work with fewer disks (reducing cost per gigabyte). Security is a measure of the number of drive failures that an array can support. Performance involves a number of different factors but the most important to assess are probably the balance between read and write operations and between transactional and sequential Input / Output (I/O):

■ Read / write balance - depending on the server role, disk activity may be predominantly reads or writes (for example, 80% reads and 20% writes). Some RAID levels have better read performance and some have better write performance.



Page 57

■ Sequential versus transactional (or random) I/O - this refers to how data on the array is typically accessed. Sequential I/O means reading or writing data at the same physical location; transactional (or random) I/O means the controller has to scan different areas of the disk or array to retrieve or write the data. In terms of database applications, transactional I/O is associated with adding and updating records while sequential I/O is more typical of querying data for reports or serving large files. The OS and user file access is also typically transactional but a media server would use predominantly sequential I/O.

RAID 0 (Striping Without Parity)

Disk striping is a technique where data is divided into blocks and spread in a fixed order among all the disks in the array. The block stripe size is often 64K but can be configured to an optimum value between 2K and 512K based on the size of the array and the type of data served (in much the same way as a cluster size can be chosen for a volume file system).

Because it provides no redundancy, this method cannot be said to be a true RAID implementation. If any disk in the set fails, all data is lost. It is used to improve performance by spreading disk I/O over multiple drives.

RAID 0 (striping) - data is spread across the array

This strategy requires between 2 and 32 hard disks. It provides the best performance when used with multiple disk controllers.

RAID 0 would never be used for live data storage but is useful for hosting read-only data (in a front-end web server or media server to serve files that are backed up elsewhere for instance) where speed is more important than redundancy.

As well as data security, RAID 0 risks availability, as the failure of one disk causes the whole array, and therefore the server, to fail. You could compensate for this by duplicating servers (clustering) but only at considerable expense.




Page 58

RAID 1 (Mirroring/Duplexing)

Mirroring requires two hard disks and a single disk controller. It takes place at the volume level and any volume, including the boot/system volume can be mirrored. Each write operation is duplicated on the second disk in the set, introducing a small performance overhead. A read operation can use either disk, boosting performance somewhat.

This strategy is the simplest way of protecting a single disk against failure. If one disk fails (degrading the array), the other takes over. There is no impact on performance during this time, so availability remains good, but the failed disk should be replaced as quickly as possible as there is no longer any redundancy. When the disk is replaced, it must be populated with data from the other disk (rebuilding). Performance while rebuilding is reduced, though RAID 1 is better than other levels in that respect and the rebuilding process is generally shorter than parity-based RAID.

RAID 1 (Mirroring) - data is written to both disks simultaneously

In terms of cost per megabyte, disk mirroring is more expensive than other forms of fault tolerance because disk space utilisation is only 50 percent. However, for peer-to-peer and modest server-based LANs, disk mirroring usually has a lower entry cost because it requires only two disks. Stripe sets with parity (RAID level 5) require three or more.

Duplexing is simply a mirrored pair with an additional disk controller on the second drive. This reduces channel traffic and potentially improves performance. Duplexing is intended to protect against controller failures as well as media failures.

RAID 1 (Duplex mirror) - an extra controller provides redundancy

server plus

Documents