cloud adoption toolkit
DESCRIPTION
Professor Ian Sommerville from The University of St Andrews discusses the Cloud Adoption Toolkit that his team has developedTRANSCRIPT
JISC meeting, December 2011 Slide 1
The Cloud Adoption ToolkitProf. Ian Sommerville
St Andrews Cloud Computing Co-laboratory
JISC meeting, December 2011 Slide 2
Cloud Adoption
Our aim is to support decision makers thinking about migrating systems to a public
cloud platform
How much might it cost?How should I price my services?
What are the risks/benefits?How will it affect my
organisation?Will it save energy?
JISC meeting, December 2011 Slide 3
Scope
• Medium-scale enterprises that currently run most services in-house
• Tens to hundreds of servers
• Incremental service migration to cloud, where services are provided on dedicated infrastructure
JISC meeting, December 2011 Slide 4
Not just a technical problem
• Concerns from staff about how the move will affect their jobs
• Concerns from users over ‘loss of control’ over ‘their’ systems
• Concerns from IT management about maintaining their organisational influence
• Concerns (real and misplaced) about security, availability and compliance
JISC meeting, December 2011 Slide 5
Cloud adoption toolkit
• Infrastructure cost modelling
• Energy modelling
• Risk and benefit analysis
• Stakeholder impact analysis
• Service pricing
JISC meeting, December 2011 Slide 6
Cost modelling
• Cost modellers are provided by major cloud providers
• Simplistic usage assumptions
• Difficult to do ‘what if’ calculations
JISC meeting, December 2011 Slide 7
Our requirements
• Cost modelling and comparison for different providers
• Variable usage patterns
• Simple ‘what if’ analysis
• Infrastructure model to allow comparison with in-house provision
JISC meeting, December 2011 Slide 88
Stand-alone prototype implemented and tested with several case-studies
Now developing web-based system
JISC meeting, December 2011 Slide 9
Supported clouds
JISC meeting, December 2011 Slide 10
Patterns
• If you simply need a service that runs all the time with a predictable, steady workload, you will not save money by moving this infrastructure to the cloud
• Cost savings from the cloud come when you have a variable workload and you provision your infrastructure to cope with periods of higher demand
• Patterns are a key feature of our tool that allow fluctuating demand to be modelled
JISC meeting, December 2011 Slide 1111
JISC meeting, December 2011 Slide 1212
JISC meeting, December 2011 Slide 13
Infrastructure description
• Analysis is based on an XML model of infrastructure AND the applications deployed to that infrastructure
• Very hard to predict overall infrastructure loading
• Easier to predict service/application use
JISC meeting, December 2011 Slide 1414
JISC meeting, December 2011 Slide 15
JISC meeting, December 2011 Slide 16
Socio-technical analysis
• Its not all about money
• Intended to assess risks and benefits of migrating systems to the cloud
• Allows complex stakeholder relationships to be considered and the impacts on these stakeholders
JISC meeting, December 2011 Slide 17
JISC meeting, December 2011 Slide 18
Case studies
Oil & Gas IT Company• A Khajeh-Hosseini, D Greenwood, I Sommerville, "Cloud Migration: A Case
Study of Migrating an Enterprise IT System to IaaS," 2010 IEEE 3rd Int. Conf. on Cloud Computing
• D Greenwood, I Sommerville, "Responsibility Modelling for Identifying Sociotechnical Threats to the Dependability of Coalitions-of-Systems", 2011 Sixth IEEE Int. Conf. on SoS Engineering
CS School in St Andrews• A Khajeh-Hosseini, D Greenwood, J W Smith, and I Sommerville, 2011, "The
Cloud Adoption Toolkit: supporting cloud adoption decisions in the enterprise." Software: Practice and Experience. Special Issue on Software Architectures and Application Development Environments for Cloud Computing, January 2011
Large media corporation and CiteSeerX
• A Khajeh-Hosseini, I Sommerville, J Bogaerts, P Teregowda. 2011. Decision Support Tools for Cloud Migration in the Enterprise. IEEE 4th Int. Conf. on Cloud Computing
JISC meeting, December 2011 Slide 19
Service pricing
• New addition to toolkit intended
• Addresses the key question for service providers
“How much should I charge for a service and what revenue can I expect”
• Price setting and revenue simulations
• How much will it cost to provision infrastructure for provided services
JISC meeting, December 2011 Slide 20
System status
• Stand-alone prototype developed and tested in a number of case studies
• Now being re-implemented as a web-based service
• Initial version scheduled for release around end-February 2012
JISC meeting, December 2011 Slide 21
Summary
• To understand the real costs of cloud usage, you need to understand and model how your applications and infrastructure are used
• Our cloud modelling toolkit allows more realistic cost simulations than the tools provided by cloud providers
• Socio-technical issues (threats, risks, benefits) are usually more significant in decision making than cost-savings
• Our toolkit is unique in integrating cost and socio-technical analysis
JISC meeting, December 2011 Slide 22
Thanks
• The development team
• Ali Khajeh Hosseini
• Derek Wang
• James Smith
• David Greenwood