Does storage utilization really matter?

Computerworld - Some colleagues and I were discussing storage (what else!) over dinner last night when one put forth a seemingly radical notion: When it comes to managing storage efficiently in enterprise environments, utilization doesn't really matter.

He was suggesting that file-system utilization represents such a small factor in overall storage consumption that it is nearly irrelevant. Was this just the scotch talking, or does he have a point?

As raw storage is configured and allocated, of course, usable capacity is lost. We know that raw storage configured in a RAID 1 or RAID 10 layout loses 50% of its usable capacity, while a RAID 5 configuration retains a significantly higher usable capacity depending on the specific layout, and there is always additional overhead in the allocation and file-system creation process.

Then data-protection policies come into play. To minimize the recovery time objective and/or the recovery point objective, multiple copies of split-mirror and replicated volumes are often needed. It is not uncommon to consume anywhere from four to 10 times the initial capacity for data-protection purposes.

Let's not ignore performance demands either. As individual disk capacities have increased, so has the challenge of providing sufficient I/O performance for high-transaction applications. Despite high-capacity disks, these applications demand placement across large numbers of spindles, occupying only the outer tracks with a large portion of each disk unallocated.

Now let's consider utilization. The argument is that after applying protection and performance policies, only about 7% to 14% of the storage is available to hold primary data. The difference between 40% and 70% utilization of this small quantity doesn't dramatically affect storage efficiency.

So, does this mean that utilization is irrelevant? Not really. Admittedly, this is a narrowly constructed scenario that may pertain only to some applications, but I have built it to illustrate a point. Tracking and driving toward utilization targets has value because it can result in more accurate storage-provisioning requests. If only half as much storage were requested, significantly less storage would be allocated considering the data-protection multiplier. Therefore, utilization remains an important metric, not so much to affect currently allocated storage, but to ensure efficiency in future storage-allocation requests.

Jim Damoulakis is chief technology officer of GlassHouse Technologies Inc., a leading provider of independent storage services. He can be reached at jimd@glasshouse.com.