Wednesday, July 22, 2009

From Backups to Scalability in the Cloud

Where do you backup your databases today? Are you using tape drives and shipping some of the backup archives offsite for disaster recovery preparedness? How much are you paying for your database backup infrastructure?

Different companies will likely have different answers to the above questions and chances are strategies will vary even within the same company depending on criticalness and security requirements of different data sets. But hopefully your company's backup and disaster recovery policy is not like that of the bookmarking site ma.gnolia.com that folded unexpectedly and lost data for thousands of users due to a data corruption issue.

The backup approach that caught my attention - backup to cloud - is being used by IDUG.org (the website of the International DB2 Users Group). IDUG recently revamped their web presence (and did a pretty good job transforming the site into a social site almost overnight). The updated site utilizes DB2 Express-C, the free version of DB2, for storing data. No matter how good the database software, there is no substitute for a good data backup strategy. Rob Williams, who helped setup the updated IDUG.org site, thinks using Amazon's Simple Storage Service (S3) is a pretty good option for storing database backups. And I agree with Rob - at 15 cents a GB per month the economics certainly make sense (unless u have terabytes of data), and certainly saves the hassle of shipping backup tapes to an offsite location. In Rob's blog on IDUG.org, he outlines how he setup DB2 backups and log files to be archived on S3 and shares his user exit script.

Ironically the now defunct ma.gnolia.com used cloud infrastructure for their operations, so it may be wise to not buy into all the cloud hype. Remember failures in the cloud can happen just as easily as on-premise, so reliance on any single infrastructure may not be wise. If you are going to be storing database backups on the cloud, you may want to do so in addition to keeping backups on-premise or another location rather than relying solely on a single cloud. Some cloud providers like Amazon allow you to create copies of your data and store them in different availability zones or regions so you can be insulated from outages in a single data center. Of course, depending on your needs (plus paranoa level and cost bearing capacity) you can utilize multiple cloud providers for your backups.

If you want to take database backups in the cloud to the next level, you could setup a database server that mirrors the data on an on-premise server. That is, a duplicate database server in the cloud ... think of the possibilities something like this could accomplish if the database server in the cloud could automatically keep in sync with changes on your on premise server. Yep, a parallel server that could also service live workloads. So you would have built-in backup capabilities, continuous high availability/failover standby, and disaster recovery option. And now imagine the possibilities if u could have more than one such database server in the cloud mirroring the same database... a neat scalability solution that could be used to distribute users/queries among multiple servers containing the same data.

If you think all this is just imaginary craziness, well think again. Or better yet - watch the free webinar: Scalability in the Cloud: Fact or Fiction - to find out how easily it can be done.

No comments:

Post a Comment

The forecast for dataville is partly cloudy.

The forecast for dataville is partly cloudy.