Computacenter

Print this page Print this Page 
Email this page Email this Page
Print this page Download Briefing
Take our reader survey Take our reader survey
Briefing Magazine 49
 

The Agile Approach
Server virtualisation is a technology that not only saves money but results in a more agile business, argues Simon Gay, consultancy practice leader at Computacenter

ICT departments are under constant pressure to reduce costs and perhaps there is no single area receiving more attention than the ever increasing quantity of Intel servers. All too frequently, a single server supports no more than a single application and many have been deployed by end-user departments away from the control of the central ICT function. In many cases, the servers are poorly managed and disorganised, with inadequate documentation and limited understanding of the build of each machine.

It’s time to adopt a more innovative approach to deploying and managing these servers. This requires a shift in thinking, away from seeing the ‘server’ as a physical entity – where one box is the mail server, another is the web server and so on — towards regarding the servers in terms of the work they do and the services they provide. This change in approach leads almost inevitably to virtualisation, where multiple, software-based servers are logically stored on a single machine. This not only addresses some of the problems ICT managers face, it introduces a number of benefits that offer significant opportunities for a business to increase its agility and responsiveness.

Rethinking resources
One of the most immediately compelling reasons for such a rethink is simply the resources that are consumed by physical servers. Very few servers are in constant demand, and even those that appear to bear heavy loads and seem to be in use continually are, in fact, often sitting idle. Even if a web server serves up a page every second, it may take only a fraction of a second to do that. So the machine’s processor is still doing nothing for much of the time. Other applications — for example, payroll or billing systems — may have long spells of inactivity or light use interrupted by bursts of activity. Overall, a physical server will usually show percentage of CPU usage in single digits. Worse still, there may be applications that are entirely redundant. And there may be others that are run just every now and then. A recent survey in one organisation showed that fewer than half of its 3,000 applications were actively in use.

Aside from the considerable hardware purchase cost, you have to house and maintain these machines. This is made significantly harder if many of them are scattered around the organisation. No wonder that the servers are rarely properly documented.

If we assume that these applications are important or even critical to your business, the problem is immediately doubled by the requirement for disaster recovery capabilities. The standard approach to disaster recovery is to replicate the production systems, which means a large amount of hardware sitting around doing nothing. With poor documentation of the live systems, your chances of getting a backup system up and running efficiently and accurately, with no loss of data or business, is greatly reduced. Virtualisation, by its very nature, incorporates many of the proven advantages of server consolidation. Centralisation of the facilities makes management, backup and recovery easier and more cost-effective. These benefits are further enhanced by the lower hardware overheads which also translate into reduced management and maintenance costs.

Flexible infrastructure
Recent research shows that developing efficient/flexible infrastructure is a primary aim for CIOs and virtualisation is a key step along this path.
The process of setting up a virtual server involves creating a configuration file. You can store that configuration until it’s needed, at which point you can recreate the server in very little time. That’s why disaster recovery is so much easier.
It’s also much more flexible because you can decide, at the time of recovery, which servers to recreate, and change your mind later if you like, with minimum disruption and maximum utilisation of the available hardware capabilities. This hugely reduces the requirement for the business to predict what it would need under such circumstances.

But that flexibility isn’t confined to fighting problems. With virtualisation, you’re able to clone existing servers, and make the applications they run available as a standard, pre-packaged system that you can roll out in a timely manner. For one customer, Computacenter reduced the deployment of certain key applications from weeks down to days. You don’t have to wait for hardware, you don’t have to have someone physically assemble it, install the operating system, apply the patches and install the application. With virtualisation, it’s already there, ready to run using the existing resources.

This gives an organisation unprecedented agility. You can react very quickly and effectively to changes in your business — such as opening a new branch, launching a new product, or responding to market changes. It also means you can set up development, prototype and proof-of-concept systems without either impacting your live production systems or investing huge amounts in computing capacity that will be redundant once the development activity ends. Virtualisation makes it cost-effective to create systems that will be needed for just a few weeks or months.
Similar advantages apply to live systems. By matching applications of complementary characteristics and running them on the same hardware, you can make the best use use of those resources — for example by marrying those resources that run heavily at night with those that run heavily during the day.

Unexpected demand
Furthermore, one of the reasons that physical servers are often so inefficient is that they must have spare capacity to deal with unexpected demand. With virtual servers, which share the available resources, you can fine-tune each server as needed. If your web site suddenly starts getting more hits than you expected, or a rush of orders means your order processing system is under heavy load, you simply reassign resources appropriately, in next-to-no time. All the time you ensure that you are making the maximum possible use of your computing resources.

Virtualisation systems have matured greatly over the past couple of years. But the real innovation here is not in the technology but in how you think about the way you use servers. Rather than being a necessary burden, virtualisation enables the server to become a valuable strategic tool in enhancing the agility and competitiveness of the business.

 
Computacenter