Virtualization and business continuity
One of the biggest technologies to hit the market over the last few years has been virtualization. Taking a virtualized approach to infrastructure, management and deployment means that organisations can consolidate the numbers of servers they are running and potentially streamline their development and testing procedures. Now companies that have implemented virtualization and taken the time to get accustomed to it are looking at how it might be used to help with their business continuity needs. From a server perspective, virtualization breaks the bond between the operating system (OS) and the underlying hardware. The physical elements of a server - motherboard, I/O, CPU, memory - are all recreated in software. This means that there can be more than one OS running simultaneously on a single physical machine: each one working within its own contained virtual environment. This allows organisations to have either multiple versions of the same OS, or different operating systems running at the same time. Organisations can run both Windows and Linux on the same server, or multiple versions of the same OS. By having all these virtual machines running at the same time, the server uses its resources more efficiently and therefore the organisation requires less physical hardware to accomplish its goals. This can be seen in the utilisation rates for servers running virtual machines compared to the ‘one OS, one application’ model that is the standard way to deploy software. A standard server will typically run at an average CPU utilisation rate of between five and fifteen percent; a server running virtual machines will provide greater efficiency and run at an average of between 60 and 70 percent. The potential of virtualization goes beyond just consolidating servers or avoiding having to buy new hardware. A virtual machine is independent of the hardware it is running on and can be moved around a company’s network to make best use of existing resources. Copies of these virtual machines can be saved off-site to protect against the effects of a server failure. Using virtualization for business continuity is still in its infancy, and consequently there are several issues to bear in mind when considering its use. As a fairly new technology, there are still additional management tools required to make controlling a virtualised infrastructure as easy as it should be. While there has been some progress on interoperability between the main virtualization vendors such as VMware, SWsoft and Xen, this process is still ongoing. A virtual machine is independent of the physical hardware that it is running on, but there is still a requirement for protection against failures within the physical server hardware. Undertaking a server consolidation project will cut the number of servers that are running but, paradoxically, increases the risk profile of the applications that are consolidated. This will therefore have to be factored into any business continuity planning. By definition, this reduction in the sheer amount of hardware means that the availability of each physical server becomes more important. Losing one physical server would only impact on one application, while the loss of a server within a virtualized environment will affect each virtual machine running on the platform. The hardware independence means that a virtual machine can be recovered far quicker than an application on a physical server can be: the virtual machine can be shifted over to a different server running the base virtualization platform, and it can be booted up straight away, but this has to be balanced against the loss of services for the length of the outage. Another area to bear in mind is that some applications do not run well inside a virtual environment: these tend to be programs that require large amounts of I/O or frequent memory access, rather than simply just CPU access. A large database, for example, can require more power and resources than can be delivered within a virtual machine environment – in this case, performance would suffer too much and a better approach would be to run the application on a physical server. From a risk management perspective, a virtual infrastructure is still reliant on its underlying hardware platform for availability. A reduction in the number of servers needed to provide an organisation’s applications can also be viewed as a multiplication of risk – if one server does go down, it will affect a far larger proportion of these applications than would be possible with only physical machines. Building a level of redundancy into the hardware platforms used is one way to protect against this risk, while having a remote data centre on standby will guard against the possibility of a site failure affecting the organisation. Looking at virtualization from a business continuity standpoint, the technology clearly has a lot to offer. The speed of recovery from failure in itself would be a massive benefit, but taken with the greater efficiencies in terms of power and resources that virtualization delivers means that any organisation should consider how it could be used within their data centres. However, virtualization is not a magic solution that will instantly solve all of an organisation’s issues in one go: careful planning and testing of the system will be required to ensure that the system delivers what is required, and that there are no issues with application performance affecting the levels of availability. Virtualization can be an important strategic part of any business continuity strategy, but taking the time to plan out the resources required by the organisation and how these will stop disasters impacting users’ ability to work is the most important function of any business continuity system. Paul Thomalla is with Stratus Technologies http://events1.unisfair.com/index.jsp?eid=249&seid=27
•Date: 9th November 2006• Region: UK/World •Type: Article •Topic: IT continuity |
