Cloud computing offers users more flexibility and reliability than traditional networking and server systems but like all technology it is not flawless. So what should companies and individuals look for when using cloud-based products in order to ensure their worst-case scenario does not end in them losing everything?
As big, fast and sturdy as hard drives get sensible users will always know to back them up regularly as they are never unbreakable. If it can be considered important to backup professional information regularly it is critical that companies have a Plan B for their server structure to ensure they – and perhaps more importantly their clients – have access to the information they need even if disaster strikes.
The idea of the cloud is to have hosting, internal servers, information and applications stored remotely in a flexible network that adapts to demand and costs less due to its scale. When a server in the cloud fails the network can immediately spread the workload elsewhere or even activate new servers to compensate. However depending on the nature of the package being used the same reliability may not exist in the event of a larger network outage.
“The cloud is very much an evolution and not a revolution so the old rules of planning really apply,” says Pól Mac Aonghusa, Manager for IBM’s Cloud Computing Centre in Dublin. “You’ve got to analyse the benefits of the cloud against the risks involved and not all offerings out there are at the right maturity level yet.”
Some cloud products on the market at the moment fail to offer the security people should demand and even the big players can fall victim to planning errors from time to time.
Google’s numerous products, which runs on a more simplified and consumer-friendly version of the cloud, had a mysterious outage only weeks ago leaving Irish users without e-mail access for hours. If a company as large as Google can see its servers go down it is clear that small businesses can too.
Of course having the “belts and braces” approach does not come cheap either and Mr. Mac Aaonghusa suggests companies should look objectively at what they need and want against what they can afford to have.
“First and foremost understand the basic cost versus value proposition; clearly it costs more for stronger security but you have to look at the value that can bring,” he says. “You should also test out for disasters. That seems like a silly thing to say but you have to know what would happen if a particular thing arose so you can be prepared for it.”
Conor O’Neill, technology consultant at Argolon and owner of LouderVoice.com, is currently in the process of switching his company’s server structure over to Amazon’s AWS system and is ensuring he has a Plan B and even C.
“I’m taking a geographically distributed approach with the main setup in Amazon Dublin and a ‘redundancy’ in Amazon East Coast US,” says Mr. O’Neill. “On top of that we’ll keep one non-Amazon setup in Softlayer in Texas in case the entire AWS system collapses.
“The idea is that if AWS Dublin goes down, we can cut over to US in a minute or two. If all of AWS goes down, we can cut over to Softlayer probably in less than an hour.”
Of course it does not have to be a failure on the server’s side that can leave cloud users stranded from their information; the problem can be local. If your internet connection goes down it does not matter how reliable your remote servers are; you will not be able to talk to them.
”We have all of our information backed up on site so we can access it even if the entire connection goes down,” says Mr. O’Neill.
However as Mr. Mac Aaonghusa points out, if you back up all of your information on a local server you may be wiping out the savings you are making by being on the cloud in the first place.
“You run the risk of duplicating your service if you run two structures like that,” he says. “Again companies should take an analytical approach and perhaps only back up the critical files they would need access to at all times.”
As an alternative to a local back up IBM have established a recovery centre in their Dublin offices, allowing users to travel there and gain access to their data. This is only really practical to people in or close to the capital but it is a novel approach nonetheless.
Conor O’Neill finds that in reality broadband outages are rare and even when they occur it is easy to circumvent them anyway.
“They have only ever happened to us once or twice and when it does it usually only lasts an hour or so,” he says. “We have 3G broadband dongles that we use a lot and they act as a good back up connection so even if your main provider goes you’re not stranded.”
What is crucial is that companies ask the right questions when they buy into a cloud service and are not complacent enough to think the cloud is immortal, even if it is resilient.
“It might be a new technology but like everything it’s not the be all and end all – things can still go wrong and you need to plan for that,” says Conor O’Neill. “For us that means having two separate servers from Amazon and another non-Amazon client just in case!”
An edited version of this article appeared in Business & Finance magazine on the 26th March 2009.