This article can also be found in the Premium Editorial Download "Storage magazine: Betting on an enterprise-level virtual tape library (VTL)."
Download it now to read this article plus other related content.
Four steps to keeping it simple.
Take information lifecycle management (ILM). It's a simple-to-understand concept--information has a lifecycle, so manage it accordingly. Duh. Don't put Napster files on the systems running financials. Easy enough, right? Yet that simple premise has spawned countless debates, violent competition and, despite all that attention, seems to have created virtually no significant impact on IT whatsoever.
So forget everything you've ever heard or considered regarding ILM, tiered storage, multilayered data and so on. Think about this instead: There are four simple stages of life for any kind of information. This rule applies no matter where information is born or how it lives; whether it's in a database, video file, a stock trade or Word document. Grasp these concepts and I promise it will change your life, as pathetic as that might sound.
Stage 1: Dynamic active online data. Data is born from an application, in some form, and lives in a state of flux where
| changes occur. This is the stage where you should focus the most energy, money and time. This is the big honking iron connected to the mainframe or the super-fast midrange NAS box. It's the top dog on the pyramid of data life--the most fault-tolerant, visible, secure and protected time of its life.
Stage 2: Persistent active online data. At some point, all dynamic data becomes non-changing or persistent. This point varies by data, application, corporate rules and your own subjective ideas, but it eventually stops changing and just "is." Yet it may still be constantly accessed. Once it becomes persistent, however, you should consider treating it differently. Why back up a non-changing object as if it were dynamic? Do you truly need to constantly replicate it to your disaster recovery site? Should you keep making copies of it for test and development? No. When the data stops changing, your processes should start changing.
Stage 3: Persistent inactive data. This data not only doesn't change, it's rarely accessed. But if we want to access it, it better show up quick. This data should still be kept online and accessible, but it doesn't need to be on super-expensive, high-performance gear, and it doesn't need to be backed up or replicated because we most likely have 11 million copies of it. You can think of this as your online archive, persistent store or infinite data repository, but you'd better stop thinking about it as you did before. This is where the majority of your data lives and it should be on your most economical infrastructure.
Stage 4: Persistent inactive offline data. This is your offsite, offline, deepest-of-deep archived copies of "Please don't ever make me read that again" media. This is the doomsday vault. Even here, however, simple process treatment will make you a happier and seemingly much more intelligent professional to your business colleagues. If you end up buying tape media only a few times for persistent data instead of every weekend, you'll cut your annual media budget by 90% ... which can translate into big, big dough.
Next month we'll talk through some real-world examples and bring this concept down to earth. In the meantime, try not to get caught up in the next overcomplicated concept du jour.
This was first published in August 2008