A P P L I C A T I O N D E V E L O P M E N T

PowerOneUse Case Review

By

Inigo Olcoz and Phil Hummel

1

Data Center Trends

In the 1960s, the average tenure for a company to remain listed in the S&P 500 was 55 years. Current projections indicate that 75 percent of the current S&P 500 member companies will be replaced in the next three decades. That dramatic turnover is due to many factors, including globalization and demographics. Superior management of data and shrewd investments in information technology (IT) are common traits of organizations experiencing rapid growth share compared to those that are in decline. Data is essential to enterprises in the digital era and IT moves the data to where it can best support organizational growth.

For decades, many organizations benefited from the improved capabilities and lower prices predicted by Moore’s Law. However, as the growth rate of raw silicon power begins to level off, the ability to do more with less will depend on systems integration and automation. While the cost of equivalent technology has declined year over year for decades, operating costs—in particular the cost of labor—have increased steadily. The need to integrate and automate is clear.

Ten years ago, it was projected that all data center operations would be managed by a few vendors hosting cloud-based services by contract. The assumption was that efficiencies of scale would drive automation and that the cost advantages of standardization would eliminate the motivation to own and operate private data centers. In fact, in the past ten years have seen growth in public/contract clouds as well as in on-premises private clouds and traditional IT models. We now realize that private data centers are here for the long run, and that large investments in ongoing modernization and expansion are fueling demand for systems that have better integration and more automation out-of-the-box.

2

• Delivered as a turnkey system that provides fully

automated VMware business outcomes

• Based on top-performing Dell EMC components:

PowerMax storage, PowerEdge MX7000 servers,

and PowerSwitch networking

• Designed and built to address the customer’s

ongoing investment in data center automation

• Aimed to accommodate the increasing technology

consumption required for achieving digital

transformation objectives

• Designed to be massively scalable to support

thousands of servers and multi-petabytes of

storage

• Allocation of SAN Fabric and storage

• Allocation of compute resources through a

dynamic, zero-midplane server infrastructure

• Administration of network resources by using

near-zero-touch PowerFabric administration

• Infrastructure initialization, configuration, and

provisioning

• Operational analytics and monitoring

• Inventory management

A key initiative of digital transformation is the optimization of the datacenter infrastructure layer. The PowerOne System is a new Dell EMC converged infrastructure (CI) offering that is:

PowerOne Controller is a key component of the PowerOne System, providing increased levels of performance by automating:

PowerOne Systems are engineered and assembled by a single vendor, providing a seamless customer experience that is fully supported by Dell EMC.

PowerOne Controller reduces the time required to develop and deploy products, thereby reducing the infrastructure’s time-to-value. Because the PowerOne System is designed according to VMware Validated Design (VVD)

Through the effective use of automation, PowerOne Controller defines new operational paradigms in:

Growing and Uncertain Development Environments

Bursty Test Environments

High Latency Production Environments

Collapse

PowerOne Consolidates

• Environments

• Management

• Infrastructure

PowerOne Advantages

3

best practices, it can easily fit in with the customer’s migration to the hybrid cloud and become a supporting element of a Dell Technologies Cloud (DTC) implementation. Using PowerOne Controller, deploying and managing VMware based workload domains has never been easier.

To respond to the ever-present market demand for cost optimization, the PowerOne System is built around asymmetrical scaling principles. Asymmetrical scaling allows for infrastructure growth just in the required layer (compute, memory, or storage), independently of the others. PowerOne offers the best alternative for business workload scenarios where there is a heavy demand just for storage or just for compute, avoiding the huge infrastructure costs associated with classic homogenous scaling. The resulting infrastructure is less complex and provides significant software and licensing savings by allowing infrastructure resources to scale only as required.

PowerOne management is based on its RESTful API and the API’s front-end UI, PowerOne Navigator. PowerOne Navigator provides simple wizard-based operations that enable the customer to initialize, configure, provision, manage,

and expand their infrastructure.

PowerOne Navigator showing PowerOne components and status

4

PowerOne Navigator also lets you automate the life-cycle management of your IT infrastructure to simplify data center operations, reduce infrastructure cost of ownership, and deliver a semi-autonomous management and operating experience.

PowerOne’s asymmetrical scaling capabilities make it the perfect choice for consolidating data center workloads and for hosting the most resource-demanding business-critical workloads. PowerOne’s PowerMax storage arrays, MX7000 servers, and PowerSwitch networking are ideally suited for hosting virtualized or bare-metal SAP, SAP HANA, or Oracle deployments, and for building high-performance SQL or Exchange clusters. PowerOne is designed to satisfy the demands

of memory-intensive, low-latency, and high-bandwidth workloads.

PowerOne Use Cases Being Driven by Digital Transformation

Providing IT services for software development teams has always been a challenge for data center

professionals. Despite the high cost, the best practice of maintaining multiple “identical” stacks for production,

testing, and development was a standard for decades, starting with the commercial introduction of mainframe

computers. The legendary reliability of those mainframe environments required lengthy and expensive testing

and validation protocols on other systems before a new program was accepted into the production environment.

This strict separation of hardware resources was based on the standard “waterfall” software development stages

of dev/test/prod. It was carried forward through the move to client server computing and into the first decade of

Internet scale web application development.

When VMware revolutionized server virtualization for the x86 platform, it reduced the cost of data center

operations for many use cases, including software development. Virtual machines were more efficient at

providing isolation between compute resources for multiple software development environments because

resources no longer had to be allocated in increments of full servers. However, reaping the benefits of

virtualization for operations still required many manual processes for building and expanding vSphere clusters.

Also, the nearly complete lack of virtualized networking and storage available during the initial period of rapid

server virtualization growth made automating full developer stacks impossible.

To address this situation, PowerOne provides automation during initial system configuration prior to installing

the virtualization layer (Launch Assist), and during ongoing operations (Life Cycle Assist and Expansion Assist).

PowerOne also provides full stack automation of the network, compute, and storage layers. The tight integration

between the PowerOne Controller and each of the underlying Dell EMC developed element managers is one of

benefits of investing in hardware and software from the only tier-one vendor focused on full stack automation.

Use Case – PowerOne and Application Development

5

Early server virtualization efforts faced skepticism about the ability of a virtualized infrastructure to provide the

performance required for production environments. It was previously believed that at least one identical physical

environment as closely matched to production as possible was required for integration testing prior to release

to production. This resulted in significant costs and complexities that virtualization could eliminate. Over the

last ten years, even the most demanding workloads have successfully moved to a fully virtualized environment.

This industry shift confirms that a converged system such as PowerOne can provide an infrastructure capable

of supporting all your software development and production environments. By enabling full stack automation for

networking, compute, and storage, PowerOne eliminates additional cost and complexity from your data center.

The next revolution to hit the data center was the publication of the “Manifesto for Agile Software Development”,

which challenged the notion that the following two goals were incompatible:

• Maintaining high reliability for production environments

• Quickly delivering new and updated software to end-users

As many of the largest and most profitable online brands, including Netflix, Amazon.com, Facebook, and

LinkedIn, began publicly touting their successful implementation of new software developed based on the “agile”

methodology, it became more difficult to support the policy of maintaining multiple duplicate environments for

dev, test, and production. The agile methodology includes best practices for designing and developing software,

as well as guidance for better collaboration between data center infrastructure operators and the application

developers who rely on them. Since its introduction and maturation, agile/cloud native software development

methods have experienced widespread adoption.

The progressive shift away from traditional waterfall software development to agile methods for developing

microservice architectures is at the heart of the cloud-native applications movement and has remained among

the hottest “buzzwords” at software development conferences for more than ten years. Case studies of multiple

enterprises have shown that, for applications, high reliability and faster time to delivery are compatible under the

right circumstances. The design philosophy of “do one thing but do it very well” was the organizational mantra

that allowed service-oriented architecture (SOA) to evolve into a much more consistent and robust way to design

and develop software. Microservices architecture has long been popular because of its language, operating

system, and code management platform independence. This explains its widespread adoption among so many

different types and styles of organizations. All of these organizations have similar needs for platform automation.

This underscores the broad need for PowerOne.

The two most important practices to come out of the agile software methodology are:

• The ease of moving quickly from stating the problem to developing prototype code

• The need to test various design alternatives through prototype “competitions”

Rapid prototyping usually entails creating a significant number of environments that are only used for days or

weeks before being abandoned. For organizations that lack essential data center automation, this “fail fast”

approach creates an out-of-control time demand on IT operations and support staff. Numerous people will need

6

to perform many hours of manual configuration tasks, and carefully coordinate them with other groups, while

more service tickets are coming in with requests to delete configurations and release resources. The tension

between resource requestors and configuration implementors is only made worse when teams that are engaged

in agile software development simultaneously request multiple environments for a single project to compare

advantages and disadvantages of competing designs through parallel prototype development.

It is a worthy goal to allow software developers to focus on their area of expertise. However, organizations

need to recognize that agile methods are only successful when data center operations staff are closely involved

with the project. Cooperation between developers and operators is needed to drive decisions regarding the

allocation and configuration of underlying equipment to support software developers. Having too many people

on the team, each with a narrow area of responsibility, increases communication overhead and slows down the

development process.

PowerOne gives IT operations and support teams a single tool, and in many cases a single-person approach,

to deploying environments that support agile software development initiatives. The PowerOne Controller

is the only tool needed to create and configure clusters that are isolated from other key stakeholders (such

as production operators) who are using the system. One or more IT generalists can work in parallel and

independently to configure full stack environments without the need to open tickets for support from dedicated

network, storage, or compute subject matter experts. The reduced demand on the most expensive SMEs opens

the opportunity to invest in more high value data center activities, including planning, support escalations, and

new solutions development.

7

Conclusions

Many of the most widely publicized agile software development case studies come from the successful Internet players who exhibit both high revenue and high profit growth. They have been able to invest substantially in the development of automation tooling that supports the rapid deployment of new software services. Many also contribute parts of their internal development efforts to the open-source software community, but never a complete solution. This leaves more traditional organizations and enterprises to choose between developing a complete solution or gluing together pieces and parts from the open source community to automate their agile software development initiatives. Without automating infrastructure configuration and management together with code management, agile software development is not agile.

Software developers working in environments that lack important tooling for IT automation often experience extended periods of downtime waiting for resources. Jump-starting a new development project can require creating numerous service request tickets for compute, storage, networking configuration, security credentials, and others – each of which may generate several additional tickets for supporting resource(s). A storage administrator may need networking configuration support that is separate and independent from what the developers have requested for client access to the application or specialized port configurations between layers in a multi-layer service design. Completion of these complex and interdependent sets of service requests can easily lead to missed service level agreements if any one resource misses a deadline for a task that must be completed before other teams can do their part.

Software development teams have tried to eliminate delays caused by a lack of infrastructure automation by moving to a contract cloud provider. Though these teams have seen some improvements in generating proof-of-concepts or in creating simple function-as-a-service offerings, the skills required to configure networking plus manage multi-level compute and storage assets for modern microservices remains challenging in all cloud environments that lack true multi-function automation.