Go Virtual to become Agile

The flashing blade

The hardware for such datacentres is, for the moment, most commonly a 2- or 4-processor rack-mounted server, sometime called a `brick’. The coming thing, however, is the Blade server, a thinner unit that shares resources such as power supplies, communications and even storage. These can be densely packed together and come with a racking system that provides the power supplies, connectivity and other required services. As well as processor Blades, other types are now becoming available for managing specific services, such as networks, attached storage and the like.

The key advantage is that it is a simple task to expand such an environment by adding more Blades. So there are now two possible approaches to scaling the system. One is what might be called the "permanent" approach, adding more hardware resources in the form of more Blade servers, which can be achieved without halting the datacentre’s operations. The other is the purely "virtual" approach, where the flexible nature of consolidated hardware and virtualisation technologies allows the available resources to be reassigned as and when required.

The other key component in advancing virtualisation is the necessary management software, which monitors and controls the operations of the individual servers, assigns workloads to them and delivers results to the appropriate recipient users or systems. In this way, the correct number of servers, together with the appropriate operating environment, can be made available at the time an application needs to run. When the task is completed, the application and operating system are uninstalled and the resource is made available to the next task. This approach has the potential to not only reduce the ongoing investments needed in hardware but also reduce the cost of managing the infrastructure, if only through the centralization of resources into a limited number of locations.

Raising standards

There are two other factors of importance with virtualisation and scalable systems these days, and these are standardization and interoperability. The fundamental goal is that, as far as possible, anything should work with anything else. For example, as the three dominant operating systems – Windows, Unix and Linux – all run native on the x86 architecture it is not surprising that processors using this are the most common hardware platform. Though other processor types are used, particularly for proprietary Unix applications, they are in the minority. When it comes to management software, there are a number of contenders, including IBM’s Tivoli and Director suites, HP’s OpenView and Microsoft’s MOM, but even here the major players now all make a point of ensuring that their management systems interoperate with the rivals.

So virtualisation has the potential to create operational infrastructures that really do allow users to build far more agile and flexible business processes that can be scaled to meet both occasional and permanent increases in demand, while reducing both investment and management costs.®