Application slowdown and how to solve it

Expert panel explains how to keep your data centre running smoothly, and discusses why simply throwing more capacity at the problem doesn't work

Application slowdown in data centres cannot simply be solved by throwing more capacity at the problem, and in fact that may create further problems and increased costs.

That's one of the opinions given during a Computing web seminar today called 'The curse of slow applications'.

Rich Fenton, UK & Ireland systems engineering manager at Nimble Storage, explained that adding traditional disk-based storage to a data centre will help capacity, but is unlikely to improve application performance.

"Traditional disks are great for capacity. But from a performance perspective, they're not so great. It's very hard to solve a performance problem with traditional disks, and that's where flash comes in. Using a combination of disk and flash-based storage enables you to provide a balance between cost and performance."

Also speaking at the seminar, Sanjeevan Sivasubramaniam, project engineer at law firm Stephenson Harwood LLP, explained that simply adding storage brings other overheads.

"More disks means more management and takes up more space, which means more money."

Fenton explained that by using telemetry tools in the infrastructure data centre managers can get a better understanding of how applications and the underpinning hardware is performing.

"Using telemetry from the infrastructure helps to pinpoint problems, so you don't necessarily have to go through an expensive upgrade every time you see a problem."

Sivasubramaniam added that this telemetry has helped his firm to identify problems before they arise.

"The main advantage with this predictive analysis for us is seeing the capacity we need and when it will run out. When the data centre reaches 80 per cent capacity it starts to run slower. The telemetry gives you an idea of when you need to start thinking about adding more capacity, so you solve problems before they arise," said Sivasubramaniam.

"It gets round the problem of the infrastructure guys saying it's the application's fault, and the application guys saying it's a problem with the infrastructure," explained James Elmer, director of information technology at Stephenson Harwood.