It wasn’t that many years ago that “snapshotting vs. backup” had the same kind of fervor in IT debates that some American political discussions have today, or religion, or any other media-infused divisive issue. Sidenote: my favorite of these incite-ful battlegrounds is still to enter a filled room and shout “Tape is Dead” (which it isn’t, but the ensuing conversations are worth the fisticuffs).
Imagine a dialog between two ardent IT experts, each trying to convince an application owner to use their data protection capability:
Storage Diva: Use snapshots ... because they are fast … and they don’t take server CPU
Application Owner: They are fast, but your storage is too far (logically) from my application, and doesn’t understand what the application needs before you make that restoration point. Oh, and I can’t afford to keep snapshots around for weeks or months.
Backup Dude: Do backups … they understand applications via agents … and they can retain your data for years on tape.
Application Owner: You do understand applications, but your restore isn’t near as fast. And your agents are okay, but they consume CPU (sometimes lots of it).
IT Manager (and the boss of Backup Dude and Storage Diva): Both of you go back to your corners. We can use both methods for what each is best at. App Owner, how does that sound?
Application Owner: That’s what I hoped you would say.
Conversations like how this one ended is one reason that you see so many data protection companies using superhero caricatures to emulate the backup IT Pro. Yes, you can have your cake and eat it too.
The key to remember is to start with what kinds of recoveries that you need, and then choose your data protection solution(s). I’m not a golfer but I imagine that most would tell you to figure out how far the ball should go and then pick the club; instead of using the same club for everything because that is what you are used to (and then just hit it lighter or harder).
There are easily a half-dozen different kinds of restoration agility scenarios out there, from fast rollback of production data, to instaneous VM booting, to long distance resumption of service – and for each one, there are a variety of tools that can enable that outcome. The trick is to minimize the number of tools without sacrificing ones’ recovery agility. In some cases, that might mean a single toolset that can do everything. More often, it means smarter leveraging of storage solution capabilities that are integrated and managed from the same management tool that manages your backups; possibly with incremental integration with your hypervisor or other infrastructure management lens.
First, decide how you need to recover – and then, pick your protection methods – so that you can look for vendor(s) that can either deliver the majority in one, or at least unify the management so that you have a unified lens to manage your unified data protection (and recovery) strategy.
As always, thanks for reading.