Data centre decisions: past, present and future
The cloud and virtualisation are changing the data centre playing field, but the technology is not necessarily the most important aspect to consider. Reliability, support, user experience and legal issues remain critical.
The move to the cloud, virtualisation, power provision, efficiency, latency, monitoring and infrastructure management are just some of the challenges that face data centre providers and users in 2014. According to a report issued earlier this year by Emerson, in 2007 the top three concerns were heat density, power density and energy efficiency. In 2013, it had become monitoring and infrastructure management, energy efficiency and heat density - a reflection of the move to virtualisation and the cloud.
“Australia and New Zealand have been very quick to move to a virtualised environment, far in advance of the US,” says Michele Caminos, a Gartner managing vice president based in Wellington, New Zealand. “And now they’re looking at what do we do with this virtualised environment - do we move to the cloud now, what do we do with our data centres?
“A lot of businesses are giving it back, or consolidating,” says Caminos. “And it will progress down that path as we continue to go to the cloud, as they obviously don’t need infrastructure housed within their own walls.”
To get a better idea of what data centre users see as the challenges in the current environment, we asked a number of them for their views.
Reliability and support
Australian IT consultancy IMTG focuses on providing e-learning platforms, CRM as a platform, optimisation and web applications through cloud. It has gone through the transition of handing off its data centre to an external provider.
“When we started the business, we started building our own infrastructure,” says Makis Marmaridis, IMTG’s managing director. “Nearly 15 years ago when we started there was hardly anything to speak of in the way of what you can do today. You could hire the rack space and put your own servers in or run your own network, and try to do everything yourself. And we did that for a long, long time. And that’s because we had to - we had no other choice. We had to provide reliable service and we had to have control over the environment. So we had to maintain all that overhead just for us to be able to deliver those applications at the level at which we needed them to be delivered.
“I knew we had to get out of running our own infrastructure very early on. And every time I had to sign off on a new server in the data centre, I would remind myself that this was not the way to do it long term,” adds Marmaridis. “When we decided we were going to move off our own equipment, we thought it would take us about two-and-a-half years to do the switch. It took us just over 12 months in the end.”
Marmaridis says there were a number of factors that had to be in place before he felt comfortable making the switch, but prime among them were support and reliability. “We have to be certain the equipment is absolutely bulletproof. That’s what we were waiting for. The technology was there to some degree, but there was not enough support - the support was not close enough to what we needed.”
That’s changed now, he says. Putting a ticket through and having to wait 12 hours for a response is a thing of the past.
User experience is the key
“We have around 150,000 customers globally, we send roughly 1.2 billion messages per month and we see between 25 and 30% growth in our message volume every year,” says Cameron Newman, head of operations at Sydney-based global email marketing firm Campaign Monitor.
“When [the founders] started this business, they were always focused on the North American market, so from day one all of our infrastructure has been in North America to provide the best experience to our end customers,” adds Newman. “So we’ve gone through a whole range of providers in North America.”
What’s top of the list when considering data centre services? “At the top of our list is user experience, [which is] performance related. Secondary to that would probably be support. The user experience is paramount to everything we do at Campaign Monitor; making sure that when you’re in the application the pages load very quickly and are snappy.”
“It’s a pretty important part of your reputation, that you’re always available,” says Newman. “We’ve seen some pretty ugly things happen when providers can’t follow through with their service. It’s obvious that having downtime is not something [the founders of Campaign Monitor] are willing to have to go back to customers to explain.”
And what about specifying where your data centres are located? “Yes, we do specify that. We even get down to the detail of specifications of network connectivity to their infrastructure so that we can optimise the performance all the way down to our edge server that’s rendering a web page.”
A shift in thinking
Sydney-based company Global Storage is both a consumer and a provider of data centre services. “What we’re seeing at the moment is a real shift in thinking on data centre strategy,” says David Duncan, CEO. “And that's come about because cloud is changing the focus of technology leaders and CIOs from technology relationships to service relationships.
“So rather than deploying in their own private data centre, we’re seeing a big trend for our customers particularly deploying in our virtual private cloud,” adds Duncan. “So that deployment mechanism means that their data usage is decreasing; if anything they’re cutting back on the data centre space that they’re using within the traditional providers.
“The big change that we're seeing is that public cloud when it first came in was more tailored to web services, maybe a bit of CRM and a few other different things. But the real take-up in Australia isn’t around these traditional public cloud offerings - it’s around virtual private cloud offerings.”
And where is that shift in thinking leading? “The shift is different from what we’ve seen before in terms of where the market’s at. Right now, because the way we consume technology is changing, and it’s changing quickly - we’re going to technology being consumed as a service,” says Duncan.
Commercial and legal challenges
The Tatts Group provides services in the gaming and entertainment industry. And it is a huge operation, with 10,000 physical sites, a large telephone betting operation, an online channel and around 250,000 poker machines.
“We believe it’s the largest network in the country; we can’t find anybody who’s bigger. It’s basically Coles plus Australia Post plus 20%” says Matt Maw, the company’s CTO.
“Tatts Group was formed through an amalgamation of a number of entities, and each time we’ve acquired those companies, we’ve effectively acquired - loosely - two data centres. So at worst we had three data centres in Queensland, two in Sydney, two in Victoria, one in Adelaide, one in Hobart, one in Malta and one in Milton Keynes in the UK,” says Maw. “We’ve steadily, over the last 4 or 5 years, consolidated and shut down data centre after data centre across the country and have now successfully consolidated ourselves into two data centres in Brisbane, which we both own and control.”
“We’re facing a situation where an existing building is going to be sold, so I’m now faced with the quandary of needing to shift a data centre, and we’re having some really interesting conversations at the moment as to whether or not the ownership of our own data centre is a strategic asset to us or is a cost of business,” adds Maw. “If it’s a cost of doing business, then there’s absolutely no point in continuing to operate it ourselves. So it’s really a strategic question for us as an organisation.”
One of Tatts’s biggest challenges came during Super Tuesday a couple of years ago, when an Oz Lotto $100m super draw coincided with the Melbourne Cup. “We maxed out at about 5500 transactions per second through the system,” says Maw.
How do you handle such a load? Is it a simple matter of bursting into the cloud when you need more capacity?
“We’re a little bit different in that we basically are a technology company,” says Maw. “We have no redundancy, in that if our systems go down, there are no manual processes or procedures to back them up, unlike an airline that can manually board a plane. For us, the horse race jumps, the lotto finishes, the time ticks over regardless of whether our system is up or not.
“The challenge for that is not at the technical level, it’s at a commercial and legal level that says, ‘What happens if there’s an outage or an issue?’ When you’re running 5500 transactions per second, 30 seconds matters,” says Maw. “Finding a legal contract that you can sign with any provider that says they can even identify they’ve got a problem within 30 seconds let alone begin to rectify the situation or pay out on that SLA [is difficult].
Even a momentary electrical interruption can spell disaster for your business, so it pays to not...
The old adage that you can't manage what you can't measure is nowhere more apt than in...
Looking to delay a decision on cloud, or to simply save a buck, organisations are holding onto...