By Arthur Shih
“My data is in the cloud, why do I need to worry about disaster recovery or backups?”
It’s a question that I get asked a lot.
I love the concept of cloud computing. Period. I love the fact that I can access my information at any time, from anywhere, on any device. I especially love the fact that I just have to consume the services and the content, without really worrying about the underlying infrastructure and what it does. What concerns me, though, is how people often think of it as a silver bullet – once their data is in the cloud, everything becomes someone else’s problem, including disaster recovery.
A true disaster isn’t when your service is down. A disaster is when your service is down, but a competitor’s service is still running. It’s times like these that potential customers move from one service to another, and more often than not, they never come back.
When you buy into a pure Infrastructure-as-a-Service cloud environment (like AWS or Rackspace), you are purchasing computational and storage resources to use as required. The beauty of it is you can buy it when you need it and have it immediately. What a lot of people forget, however, is that this fancy IaaS cloud you have just purchased still runs on physical hardware, and is stored in physical locations. As much as virtualisation and replication have lessened the impact of hardware failures, they have not removed them completely. A disaster may still strike and outages will happen. Think Hurricane Sandy. If your IaaS was being provided out of a datacentre located within the areas that lost power, then your service provider most likely would have had a failure, and your systems are down. And in the case of Hurricane Sandy, you have no idea when it will be back up again – unless you have a disaster recovery plan.
There are of course other events that may cause outages –Amazon have had many well-publicised failures that have nothing to do with natural disasters. The main thing to understand is that it’s not a matter of if it will fail, but when it fails.
“But I’m in the cloud, so isn’t everything load-balanced and copied across geographies automatically?”
That’s the second question I get asked after discussing the first point above. In some cases, yes, your data can be stored in multiple locations, but it is something you need to actively purchase on top of your original cloud service.
When thinking about disaster recovery, there are four key questions in my opinion that you need to answer:
- How long can I survive without my IT Systems?
- Once my systems are running again, how much information can I afford to have lost compared to the point in time when the system went down?
- Which systems are really the ones that I need up and running?
- How do my users then access these systems?
Also, never forget the human element of a disaster. You may have all the best disaster recovery plan in the world, but in the event of a disaster, it’s guaranteed that your staff will be more worried about getting home to check on their families than how to respond to customers’ emails.
Arthur Shih is Datacom’s Cloud Solutions Manager.