Capacity Planning in a Recession

Capacity Planning in a Recession

How do you get the scale of IT investment right during a recession? Most companies will agree that at least some of their IT projects are critical to their survival and continued competitiveness however tough the market is. These projects are likely to be the ones that can be linked directly to increased revenue or a decrease in costs in the very short term.

However capacity planning even for these essential investments is a difficult art. In the good times systems tend to be "built for success" read large amounts of unused hardware based on wildly optimistic usage projections. VMware estimate that servers typically operate at around 5-15% utilisation. This is a waste of valuable budget that could be spent elsewhere.

In more challenging economic times there is the temptation to cut back and provide a "leaner architecture". Like "Just-In-Time" inventory management this can save considerable money but can be an embarrassing failure if usage suddenly increases. IT professionals often push against this model correctly pointing out the false economy of not providing enough head room.

The widely accepted solution to this issue has been virtualisation. Multiple virtual servers share the same hardware and can be scaled up and down on demand. Wasted capacity can be used by another server that needs it and an unforeseen increase in usage can be handled by rapidly increasing the resources available to the virtual server that needs it.

At the individual server level that's great but the organisation still has to plan how much capacity to build into their virtualisation platform. For an enterprise level product like VMware that requires shared storage in addition to the servers the costs will likely be several hundred thousand dollars.

Businesses are increasingly turning to virtual servers hosted by a third party, typically a server hosting specialist, at least for some of their capacity. Swapping capital expenditure for a monthly payment has several advantages. Firstly virtual server contracts usually run on a month to month basis. This means the server can be turned off and you can stop paying if the project fails, rather than being left with sunk costs in hardware. You can generally start a lot smaller than you would in house counting on the service provider to keep capacity available (for all their customers) at no additional charge. Virtual servers are also uniquely easy to migrate as they are hardware independent, which means that a future consolidation project to bring all your servers in house should be relatively simple.

Contracting servers from a server hosting company has other benefits too points out Neil Webster from Web Drive. More locations means more resiliency in a disaster. If your IT team need to add physical capacity to an in house infrastructure, then it will be a lot quicker with a hosting company. The average time to deploy a virtual server is hours, rather than days or weeks. Finally not having your IT team focused on hardware means they can be focused on that revenue generating change you've been waiting for.

It's not suitable for every application, admits Neil. Some applications are impacted by the user not being local to the server. "Web Drive have been able to help most companies who have come to us to host applications", explains Neil, "there are a range of solutions that we have put in place to make sure the user doesn't notice a difference".

Not a solution for everyone, but a more flexible way of managing server capacity more suitable for an uncertain world.

ENDS

© Scoop Media