Of all the unusual places to put a data centre, the middle of a zoo must rank among the most bizarre. But for the Zoological Society of London (ZSL), which operates London Zoo, it is the obvious location.
Until August 2010, the Society’s that data centre was bursting at the seams. “I had a large number of physical servers in a small, not-very-purpose-built space, and it was full,” explains head of ICT Nick Napier. “I was struggling with air conditioning; I was struggling with managing it all; and my biggest worry was our ability to recover it all in the event of a significant disaster. I dread to think what that would have meant.”
It is only in the last few years that the Society has been so dependent on IT that an outage would have been catastrophic. “Until two or three years ago, [a failure] would have been pretty bad but it wouldn’t have caused the business to grind to a halt – we could have found ways to run our ticketing systems manually for a while,” Napier explains.
“But now rely so much on IT now that I was very conscious that we were in a position that we needed to do some work,” he says.
The solution to Napier’s management, cooling and business continuity concerns was virtualisation. Last year, the ZSL replaced 37 physical servers with three HP blades running VMware.
Doing so has cut cooling costs and CO2 emissions, Napier says, and has removed the need to move to a new, purpose-built building – costing £200,000 – that was at one point being discussed. Most importantly for Napier, it has greatly improved the Society’s resilience.
“We’ve put in a similar but slightly smaller setup at our sister site, Whipsnade Zoo,” he explains. “Using replication on the NetApp storage infrastructure and the VMWare site recovery tools, we now replicate all our data and servers across the two sites,” Napier explains.
“In the event of losing the London site, I can bring up all the servers at Whipsnade and run the entire organisation from there. And assuming I’ve got the right communications linking me back to London, then I can service that site too.”
Before implementing the virtual environment, Napier and his team did not have the ability to test new software before it went live. “An upgrade would have meant we had to shut everything down, back it all up, do the upgrade, hope it works and have a rollback plan,” explains.
“Now, we can actually do it all the test environment and have confidence that there aren’t many holes we’re going to fall through. We’ve got a level of control that, for a team our size, we wouldn’t really have had before.”
The nature of Society’s ticketing systems and the poor connectivity at the Whipsnade site preclude a wholesale move to cloud computing, but Napier says that he is considering moving non-core elements of the IT infrastructure, such as the Microsoft Exchange server, into the cloud in future.
“In the event of an outage, I’d be more concerned about bringing back ticketing and other applications,” he says. “If I can run [email] in the cloud it provides a way for the organisation to communicate without me having to worry about it."