Change is the only constant...

Originally posted on “Databases at CERN” blog

Most of you for sure know, that ability to restore data in case of failure is a primary skill for each DBA. You should always be able to restore and recover data you’re responsible for. This is an axiom. To be sure, that you’re able to do it, you should test it on regular basis. There is of course possibility to use some Oracle features, like BACKUP ... VALIDATE or RESTORE ... VALIDATE commands, but if you want to be certain as much as possible, the only way is just to run real restore and recovery. Doing it periodically for big amount of databases is extremely tough, both because of resources needed and DBA time. That’s why one of our DBAs Ruben Gaspar Aparicio has created recovery system, which is heavily used at CERN. Good news – it is available as Open Source on Sourceforge (Recovery Platform). Since its release we’ve introduced many modifications to it, but still it could be good starting point to check the source code in order to start developing your own solution. We’re using it as a validation of our backup strategy, to run real restore and recovery every week or two for most of our databases.

Continue reading “Importance of testing yours backup strategy”

Originally posted on “Databases at CERN” blog

Physical standby databases seem to be ideal candidates for offloading backups from primary ones. Instead of “wasting” resources (unless you’re already using Active Data Guard for example), you could avoid affecting primary performance while backing up your database, especially if your storage is under heavy load even during normal (user- or application-generated) workload. So, if you’re seeking for good reasons to convince your boss/finance department/etc. that having standby database(s) is a must in your environment, ability to offload backups from primary databases would be for sure important (apart from usual ones, related to disaster recovery, etc.).

Continue reading “Backups in Data Guard environment”