Disk RAID is a way to take multiple inexpensive harddrives and to use them together to provide improved performance, redundancy and fault tolerance. Now, the basic concepts of RAID have been applied to SQL databases with Clustered JDBC (C-JDBC) [Slashdot]. The basic idea is to put a thin JDBC layer in front of any JDBC compliant database. All JDBC calls are proxied by C-JDBC and replicated to the underlying databases themselves.

The potential advantages in load balancing, redundancy and fault tolerance are clear. For instance, queries can be handled by any single box, thus minimising the load on a per-database basis. Inserts and updates will still need to be performed on all boxes, but this could also be optimized. Plus, if any of the database systems goes down, the other boxes in the cluster will assure that the overall system keeps running, though possibly in a degraded state.

MySQL has been working to add these features and Oracle has had much of this for years. But, with C-JDBC, it is now a simple matter to slap this functionality on top of any database you like. 10:50:17 AM    comment []