When I started at my current position, database backups were handled in a lax fashion. The full backups were performed every other day. No Log or differential backups were taken at all. The backups we stored for 90 days. Roughly 90% of the databases were in simple mode.
Adjusting this was the first thing I did. However, this had problems too. When I Changed most of the databases to Full recovery mode and added in the log backups, I started to eat at the disk space reserved for backups. Then started to expand on storage from 90 days to a year. The network engineer came to me and said to me: “dude, you are using way too much space.” My reply was, this is what it takes to properly backup 30 servers of 20TBs of data.
As a side note., most of the restores we do are on 4-5 of the 30 servers, and we usually restore next to the database, for example. Database “Sample” I would restore the database name of “Sample_”+<date of backup>+”_”+<requestor>. I do this, so the requestor knows what file I restored. The requester is in the name. This way, I can remember who to bug about removing it.
Time is not usually a critical issue on restores.
We opted to use backup deduplication software. Yes, is it slower to restore. Backups are about the same time. But I can have multiple lockers of backups and durations on backup.
I have a daily backup locker that holds backups for 25 hours. Weekly that stores them for five weeks. Monthly that keeps them for thirteen months. And long-term is for storing backups for multiple years with variable durations of expire days. The long-term backups are executed manually, where the others are all automated.
The network engineer is no longer complaining to me about space. We store some backups locally for 90 days and the rest on the cloud as the long-term and monthly backups. I can backup logs every 5-30 minutes, differentials every 1-4 hours, depending on the system. With the database’s average size, the request to restore is only delayed an extra 5-10 minutes. But think about what happened to my RTO (Recovery Time Objective). I went from 2+ days down to minutes. Better yet, no rebuilding of the servers, no advanced products like enterprise edition. The deduplication solution was part of another purchase.
If we have a critical transactional system that I needed backups stored locally and the 10 minutes made a difference, I would push to have that as a system with better control of the backups.
How are you handling backups? Are you managing the space required to do it the correct way? I can help answer these questions. Let’s get your system to work within your RTO
