I think you'll find the various backup vendors proclaiming they have "best practices". And in fact, for their technology, they probably do, but that doesn't mean it is a "best fit" for all businesses. The issues of large-scale backup and recovery are difficult and their array of variables is enormous. I'm not sure there is a universally applicable best practice for backup.
That said, I believe strongly that a backup best practice has to be centered around two general items - the creation and maintenance of backup metadata and media management. Backup metadata is what allows you to find data objects easily for restore. The metadata system is enormous and must be periodically cleaned to keep backup performance in line. Media management is critical because media is where the backup data resides. The two elements: metadata and media management form the heart of any data recovery system AND IT IS THE USER'S RESPONSIBILITY TO SYNCHRONIZE THEIR OPERATIONS. Apply your rigor to these functions and most everything else will fall in line.
Mirroring and replication and snapshot technology all could have a profound impact on backup operations, but not necessarily backup best practices. If you are quite certain that your mirrored site will not take a hit at the same time as your primary site, then it may be plausible to run full backups less often. You run the risk of having longer restores and needing more tapes to restore from if a dual catastrophe does occur. But if this gives you additional breathing room in your day-to-day operations, then maybe it can be done.
You have to ask, what is my level of confidence that everything won't go pathologically haywire? (Are you feeling lucky?) For starters, your mirroring network has to be extremely solid. If you assume the mirror is reliable, are you prepared to deal with a sequence of events where the mirroring system fails and then a dual-site catastrophe occurs? If your backup metadata and media management are rigorous and all data is accounted for, then you are OK. If not, then mirroring doesn't change a thing. In other words, if you have a backup best practice working it may allow you to adopt a different operating strategy.
I prefer differential backups if they can be done. They make restore much easier to deal with. Unfortunately differential backups might not be possible if the amount of changing data is large. To say that differential backup is a better practice than incremental would establish an artificial rule that could be misinterpreted by auditors and bean counter types that don't understand and have no interest in technology and operating complexity.
Editor's note: Do you agree with this expert's response? If you have more to share, post it in our Storage Networking discussion forum at http://searchstorage.discussions.techtarget.com/WebX?replyToMessage@7.uALMagHuauI^1@.ee83ce4!viewtype=convdate
This was first published in November 2001