At this autumn's Infrastructure Management World conference, half of the conference goers surveyed said they are or will employ server virtualisation. Meanwhile, a leading virtualisation vendor estimates that more than 90 per cent of virtualised and consolidated IT infrastructure are in production environments, and some organisations are realising consolidation rates of up to 30 to 1.
The potential benefits of virtualisation are driving the rush to consolidate entire data centres. Our work with customers, however, suggests there are risks to consider.
Those risks fall into four categories.
- Server failure. Large-scale consolidation may put many key processes, applications and services in the same proverbial basket. Consequently, fewer physical servers bear the workload -- and a physical failure has much more significant consequences.
- Over-provisioning. Starting consolidation without a clear picture of an application's function, workload or profile may lead to infrastructures that are out of balance and over-provisioned.
- Operational process. Many IT organisations implement proactive monitoring systems and some formally change control process, but few have advanced operational processes to manage crucial aspects of a smooth-running virtual environment.
- Service levels. Virtualisation technology requires new skills; for example, the ability to identify whether a problem originates in the physical or virtual environment. Without staff skills to address problems, service levels may suffer.
Operational process encompasses everything from application availability to service levels and problem-solving. Configuration and capacity management deserve close attention for their potential to ensure business benefits. The IT Infrastructure Library (ITIL), a widely accepted source for best practices and guidance in IT services, provides some helpful information.
Consolidated infrastructure requires more rigorous controls and operations practices to prevent unscheduled outages. Configuration management includes identifying system, hardware or software configuration; managing changes in configuration; and documenting configuration throughout product life cycle.
Unauthorised, undocumented and untested changes to consolidated servers could cause outages to many parts of the business that can't be fixed with a midmorning reboot. For virtualised environments, a configuration management process is essential to retain the benefits and cost savings of consolidation. IT pros should consider the following steps:
- Reiterate the need to follow existing change control processes.
- Document server and application configurations.
- Determine relationships and dependencies between servers and applications and other parts of the infrastructure.
- Implement tools that provide alerts when configurations are changed.
- Become familiar with the ITIL configuration management process and associated technologies, and implement them.
Since applications have unique load characteristics, physical server resources are used differently by the consolidated software they support. Capacity management spans planning, justifying and managing a system. When done properly, it helps determine if existing resources can support more consolidation and where added hardware might be needed.
Many organisations consolidate applications without full knowledge of the workload the application represents, or when peak load occurs. As a result, that load may not be balanced across physical servers, and IT staffers may over-provision services to accommodate uneven use of resources. IT pros should consider the following actions to ensure performance levels and reduced total cost of ownership:
- Understand the workload/profile of each key application and profile them with appropriate tools.
- Establish performance baselines of existing servers and applications before consolidation.
- Use management tools to help model workloads and virtualized infrastructures.
- Become familiar with and implement the ITIL capacity management process.
Virtualisation may significantly increase complexity with additional hardware, software, clustering and other failure-recovery features, as well as increased reliance on networking and storage services. Consequently, IT staffers must better understand server hardware and performance characteristics, what's needed to adequately monitor performance and capacity, and how virtualisation software leverages server resources. They can take these steps toward proficiency:
- Understand the organisation's server architecture.
- Develop proficiency with virtualisation technology in use.
- Build proficiency with Windows and third-party management and performance tools.
- Understand networking and storage technologies in use.
- Observe operations practices as defined in the ITIL.
IDC predicts that 500,000 servers will ship with virtualisation capabilities this year, rising to 1.2 billion by 2009. With a methodical approach, IT pros can plan for successful virtualisation and build a strong foundation for future use of the technology.
Christopher Burry is technology infrastructure practice director and a fellow at Avanade, a Seattle-based integrator for Microsoft technology that's a joint venture between Accenture and Microsoft. Gary Darbyshire is enterprise technology infrastructure architect at Avanade. Comments or questions can be sent to [email protected].