Ensuring that your website will be up and running during a natural disaster probably isn't an everyday priority at many companies, but Hurricane Sandy is yet another reminder that a disaster recovery plan should always be in place.
Actually, wherever you are, this should be a top business priority when you think about how many disasters you can't plan for, such as earthquakes. Not only is this important from a financial standpoint, but also in case you have a business that people might turn to for vital information and help during a catastrophic event.
For example, websites for several news organizations went down on Monday because their sites are hosted on data centers in Lower Manhattan, which is suffering from severe flooding and no power.
See also: Why it's not a good idea to put your servers in a flood zone
Chris Leigh-Currill, chief technology officer from global Infrastructure-as-a-Service provider Ospero, replied with a few tips for hosting solutions and servers in the case of a major disaster.
His ultimate message is pretty straightforward, asserting that the a plan in place ahead of time is "key to guaranteeing continuity of service in any Disaster Recovery (DR) event -- especially a natural disaster -- is to be prepared. Once the event has occurred it is too late."
Here's are Leigh-Currill's top five tips for ensuring your best changes at keeping your online presence up and running in the case of a disaster:
- First thing to do is establish which systems are critical to business continuity and concentrate your efforts on these.
- Ensure that the DR site is geographically disparate. In other words, ensure that both data centers cannot be effected by the same event. So, for example, if your production data center or cloud is based on the East Coast, ensure that DR is located on the West Coast.
- It is also vital that you ensure that your DR actually works. Regular DR tests should be conducted -- and this means more than once a year.
- Ensure that you have a enough compute capacity to run your business critical solutions and connectivity is sufficient.
- Ensure that everybody knows what to do in the event of a DR strategy. There is little point in replicating your entire compute estate if nobody knows how to access it.