This article is part two of a two-part series. Read Management guide: Data centre migration (Part 1)
Recognition that businesses are now crying out for new, energy-efficient space has led to a flood of investments in the data centre market, with traditional telcos and third parties expanding or building new facilities to tap into the new wave of data centre migration.
HP, for example, recently announced a $119 million data centre investment in Sydney’s western suburbs, while industry giant Global Switch has earmarked $200 million to expand its presence in Sydney.
Even application providers are investing: Online ERP vendor NetSuite, for one, “has spent $100 million in the last 14 years building a business application [and infrastructure to host it],” says regional managing director, Mark Troselj. “That gets you a pretty serious solution for any business.”
This trend is also being fed by the anticipated takeup of Cloud computing, which builds on past virtualisation investments to offer a high-end infrastructure with both improved reliability and significant burstiness that saves businesses from having to guess at future capacity.
Gartner’s survey, for example, found that 42 per cent of respondents planned to begin or continue investing in Cloud computing this year.
Businesses “are buying because they don’t want to dimension their IT infrastructure to their peak load,” says Aidan Tudehope, managing director for hosting with Macquarie Telecom, which has seen growing demand from the likes of SBS and Quickflix as it gears up to spend $60 million expanding its data centre presence in North Ryde, Sydney.
“Their server rooms have run out of power and cooling, and they want to dimension their data centres to their average load and use the Cloud to handle peaks and spikes,” he explains. “This started out as shared hosting, developed into managed hosting, complex hosting, and then virtualisation. Now it’s about the Cloud — and the Cloud is about virtualisation.”
More than rip and replace
If you are convinced of the benefits a new data centre could provide but still figuring out how to go about the migration, it goes without saying that you should proceed with extreme caution.
In an environment where risk-management requirements mean even small changes to server configurations often require extensive corporate review and signoff, a mass virtualisation exercise or a physical move is not just like throwing the baby out with the bathwater, but the entire contents
of the nursery. Whereas incremental migrations might have been possible when each server represented a wholly different function, the grand interconnectedness of modern data centres has dramatically changed the story.
“Everything has become mission-critical,” says Hanrahan, “and it has become very difficult to hive off just the Tier 1 applications for disaster recovery and business continuity planning. You have to look at everything, and we’re having to take a very holistic view of DR and BC compared with the way we would have done in the past.”
The complexity of existing infrastructure means you’ve got to consider whether it’s more appropriate to consolidate your existing infrastructure into the existing facility — which at least means you’re working within known parameters — or go the full hog and physically shift to a new facility.
The latter approach is fraught with risk, of course, but it does provide leeway for you to build the new environment while operating from the old one, then stagger the migration between the facilities as appropriate.
It’s also likely to be the unavoidable choice for organisations that want to exploit new power-saving designs that are too expensive and complex to retrofit into existing facilities.
Like a high-tech game of Jenga, however, either approach is likely to reveal unexpected issues as you begin pulling old infrastructure apart. “You need to be very, very careful if you’re doing migration activities within an existing data centre,” says Emerson’s Mandahl.
Watching, learning
Yet while shifting sites may be possible for larger organisations, it may be beyond the realm of possibility for smaller organisations. This was the case for ZettaGrid, a data centre operator that has recently been pushing consolidation hard throughout its data centre facility as it ramps up its presence in the emerging Cloud-computing market.
With a business model that revolves around VMware virtualisation and the support of the data centre consultancy arm of IBM, ZettaGrid had installed 40 server racks within its space in the Global Switch data centre, and has 1000 to 2000 virtual machines under management at any given time.
In recent months, ZettaGrid has been consolidating its remaining physical servers to virtual machines, working to identify servers that were prevented by compatibility issues from going virtual.
“At the end of the day, for us it came down to a few big problems: Power, and density, and power density as well,” says general manager Nicholas Power. “We’re finding more and more customers effectively wanting Cloud-based services, and we need room to expand — but it’s getting more difficult to find new racks as demand increases. So we’re trying to utilise our server racks for revenue-generating activities instead of using them for our internal infrastructure.”