Moving your disaster recovery infrastructure to the cloud can help your business save. But picture those savings in dollars, not hours and minutes.
One advantage of cloud computing is its scalability — use only what you need, when you need it. And of course, another is having quick access to remotely stored data.
Since most enterprises don’t face catastrophic data disasters very often, doesn’t that make cloud the perfect choice for DR? You won’t pay until disaster strikes (which hopefully will be never) and if it does, you’ll be back up and running in no time.
Well, sort of.
“The cost savings are pretty significant when you look at it this way,” explains Forrester Research analyst Rachel Dines. “In a disaster recovery in the cloud model, most of the time you’re only paying for storage and you only pay for the compute resources when you’re testing or having a disaster.
“In a traditional model, in some way or another, you are paying all the time for the compute resources. From the most straightforward approach, if you own two data centres yourself, if you have to buy servers to go at the other data centre, that’s an investment.”
However, Kaushik Ray, director of cloud solutions at Savvis Inc., says it’s a “misconception” that cloud DR is a simple pay-per-use model that will bring your whole system back up in a heartbeat. “That is true for your application or compute,” he says, “but there’s nothing out there that will automatically recover data or storage instantaneously or in a short period of time. And therefore, it means that you always need to have your data on your disaster recovery site.”
“You have to pay for that storage,” he says. “You have to pay for that replication that’s going on, even if you’re not using that data.”
“Those costs don’t go away with cloud. What does go away is if you had a large amount of cost on the compute side or the application layer. Then yes, that does go away.”
The costs don’t completely go away or “zero-ize,” Ray adds, but will be reduced significantly, depending on your RTO. If your recovery time objective is less than four hours, when disaster strikes “you need to have those application images parked on the disaster recovery site so that you can basically just spin it up.”
Part of the confusion over cost structure may be due to the fact that cloud disaster recovery is very much a new product.
There are a number of options now available, but most of have only come to market in the past year, says Dines, and “it’s not something that a lot of enterprises are going to want to jump into yet with both feet.”
For large enterprises that already outsource their DR, moving it to the cloud probably wouldn’t be “extremely painful,” she says. Those who have handled their disaster recovery in-house might be in for a bit more discomfort.
Dines says she uses three categories to define the market for cloud DR.“There’s do-it-yourself, there’s disaster recovery-as-a-service and there’s cloud-to-cloud disaster recovery.”
The first involves simply using cloud as a vessel for your own DR infrastructure, the second is offered more as a pre-packaged service, and the third is a way to set up fail-over across multiple cloud data centres, an option that would appeal to firms with very low tolerance for downtime or data loss, such as financial institutions.
“I’m seeing the larger enterprises being most interested in the do-it-yourself disaster recovery,” says Dines, “mainly because they feel that they have the expertise on-staff and some of the capabilities already in-house to do a lot of the orchestration and the fail-over and failback.”
In a broader sense, John Weigelt, national technology officer at Microsoft Canada Inc., says he sees cloud DR as an “area of great transformation.”
“In the past,” he says, “when you look at disaster recovery we have this vision of our main site, our primary site, perhaps a cold standby/hot standby that’s someplace geographically close so we can drive to it, but not so far away that it takes us on a plane to get us there.
“The cloud, with its fabric of computing, really changes that model because we don’t really have a hot standby/cold standby. We really have a collaborative environment that is geographically distributed.”
But Jeremy Suratt, senior solution marketing manager in Iron Mountain Inc.’s data backup and recovery section, argues that DR strategies that rely on network connectivity, whether traditional data centres or clouds, have an Achilles heel: the communal well can be poisoned.
“What happens if your hot/cold sites are corrupted by the same virus or issue that affected your primary data centre? In that case, tape is more reliable because it’s stored offline and out of reach, protecting your data from corruption.”
However it is accomplished, disaster recovery is starting to define how businesses operate. Enterprises are starting to realize that upgrading their DR infrastructure can be a way to get the jump on their competition, says Dines.
“I think more and more, companies are seeing disaster recovery as strategic differentiator, so thinking about it in terms of ‘if I have downtime, my competitors have an opportunity to seize market share from me, and vice-versa.’”
“It does actually, in certain industries, become incredibly competitive, especially looking at things like anything e-business, any kind of online retail or e-business.”
Brian Bloom is a staff writer at ComputerWorld Canada. You can find him on Google+. He covers enterprise hardware and software, information architecture and security topics.