Web Hosting and Disaster Recovery
I have already written about different DR (Disaster Recovery) Scenarios in the past. This last weekend an explosion brought down a large portion of an Data Center owned and operated by ThePlanet.com. As a result 9,000 customer servers were down – some for days. During the recovery portion a temporary generator broke – taking down some of those web servers again that were just back up and running. Fortunately no customer server was destroyed during the entire time.
If you are running any kind of online business this would be one of your worst nightmares to experience. How would your business survive with servers down for several days? I am not asking in regards to ThePlanet.com, but as if you would be a customer of ThePlanet.com being affected by this outage. Many people who rent dedicated servers only do backups to a second hard drive on the same server. If the server is down, they have no access to their backups. So, what is a good strategy to avoid disaster? Or what would you do, if the explosion would have destroyed your server and you would not have a backup somewhere else other than on your (now destroyed) secondary hard drive?
The answer is called remote backup. Here is an example of how The Web Hosting Resource Kit backups up websites. We actually do have a server with ThePlanet.com, but in a different data center. We backup all websites and databases to a secondary hard drive. From the secondary hard drive we backup to a SAN location (SAN = storage area network) inside the same data center. We also backup all data to an external storage provider in a different geographical region. You can see that we have several copies of our data in different locations.
Coming back to the beginning, how would one recover from a catastrophic event where an entire data center goes down? There are several options. Scenario #1: In this case some parts of the data center were back online fairly quick and the problem would be solved by itself. Let’s assume we were not that lucky and our server would still be down. Scenario #2: Let’s assume that the SAN was in a different part of the data center and is running. ThePlanet.com has servers available and (depending on the situation) a new server could be purchased (leased to be exact) and the data could have been restored from the SAN. A DNS change would have been needed, but overall data recovery would have been done in an acceptable time of approx. 18-24 hours. Scenario #3: The SAN is down, too and 24 hours later there is no sign of your server coming back online soon. We lease a new server with a different provider somewhere else or in a completely different data center of ThePlanet.com. Once the server is online, we copy our data from the external storage provider back onto the new server and restore our sites. We update the IP addresses for our DNS servers at the domain registry and we are backup and running again (with some minor loss of data depending on when the catastrophic event happened and when the last backup ran).
Conclusion: This is just one way to protect your online business in case disaster strikes. There are other options, but it also always depends on your financial situation how far you want to go. In our case the monthly cost is about $50 – $60 total and includes SAN storage, secondary hard drive and remote backup solution. I consider that cheap even though it is a lot of money, but it lets me sleep at night as well as it allows me to “ride out the storm” when disaster strikes. A great choice for remote backup solutions are Mozy and Box.net
PS: ThePlanet.com is open for business. Events like this can happen everywhere. I think ThePlanet.com is doing a good job in how they handle the situation. If you need a good dedicated server, I still highly recommend these guys. Check out ThePlanet.com
Leave a Reply