This is item #28 of the 32 types of storage ownership costs. This single category generate more discussion than almost any other items in the list of 32.

The basic premise is that high rates of waste (white space, unallocated, allocated but not used) represents an opportunity cost to the IT department, since capacity was purchased that is not being used. The argument is poor ROA. So the question is: does a sunk cost that can be recovered really reduce TCO?

 Idea 1: if the patterns of capacity waste can be minimized, then future capacity spending (CAPEX) can be reduced since more of what you have is utilized

Ideas 2: Wasted capacity is not a single cost of the capital that is not being used. Storage capacity also translates into other secondary costs:

• the cost of software, if purchased on a per-TB rate
• the cost of electricity and cooling to maintain the spinning (wasted) disk
• floorspace of the wasted capacity

One example that I recently worked on with a client that had 100TB of raw disk, taking up 4 frames in the data center. Aggregate allocation was about 50%, but actuall disk usage was 30% or less. I theory, the client could have been operating on 2 frames instead of 4. All the costs of those 2 frames contributed to the inefficiencies and poor ROA.

We all know that we cannot practically achieve 100% asset utilization, even with thin provisioning. So some level of white space has to exist. The idea with architecture decisions is to consider solutions that reduce or minimize capacity waste. Some ideas in this space include:

1. Data de-duplication
2. Tiered storage, to shift the waste (and cost) to a lower tier
3. Thin Provisioning
4. Storage Virtualization - to aggregate and reclaim stranded capacities
5. Stronger capacity planning, storage management

Some have told me that the cost of managing to the waste is more expensive than the wasted disk. This may be true, but be careful to consider all the costs involved before making general statements that can be mis-understood by management. Remember they tend to look at ROA as a measure of capital spending and efficiencies.

Trackback URI | Comments RSS