06/10/09 | in: Storage
The definition of tiered storage (assigning different categories of data to different categories of storage) seems like it would be a logical process that shouldn’t require anything special to make it work. Let’s face it, you don’t need telling what needs to go in the fridge rather than the freezer, it’s automatic.
Storage systems and the applications they support ought to be able to do this but to date very few storage systems support volumes or LUNs spanning tiers of storage. If you think about it why keep all your data spinning at 15,000rpm if you don’t need to access it?
The concept of migrating volumes across storage tiers is well established but it doesn’t really address the true need – volumes are usually made up of in-active and active data in an 80/20 split, so in most cases it make no sense to move an entire volume.
While I don’t have the answer – yet, there are vendors who can deliver these features and I believe this could be the next disruptive technology to hit the market. File level migration is relatively easy using an HSM solution but block based migration is a much more complicated concept that will require a complete re-think for most storage vendors.