Storage TCO Elements #13 and #14
by David Merrill on Oct 26, 2006
It has been a few weeks since I paused on the countdown of 32 types of storage ownership costs. A long trip to Asia, changing of Blog software and other work commitments has delayed by return to this topic. We will continue with Items #13 and #14. These next one has to do with rated availability of parts of the storage sub-system.
#13 – Storage Sub-system availability – Similar in definition to the data path availability item #12, each storage system has an inherent rating relative to the total system. Cache, controllers, drive type, power supplies, all add up to the rating of the storage array. Not all storage arrays are created equal. There is a correlation to the ruggedness, SPOF, or other design features to the purchase price.
Often storage planners can be seduced by ultra-low price disk, but there could be an availability or performance price if the systems do operate to meet the business needs or the data access needs of the environment. There are a lot of vendors (very large to very small) that offer storage at multiple price-points. Remember that price does not equal cost, and a low cost solution that has poor availability can negatively impact business operations and data availability.
#14 – Data Center Floor Space – The space required to house storage infrastructure costs money. This topic is for the square footage (meters) needed, another topic will cover A/C, power, UPS, and other environmental costs. The idea is that if data center floor space required to house storage infrastructure can be reduced, OPEX costs can be reduced. Here is the logic:
- Saving space, or reclaiming space now can avoid or delay building out the data center in the event of high growth
- Reduced space often is accompanied by reduction in AC and power
- Newer storage arrays have more-TB-per-square meter than older systems
- Efforts to increase utilization within the existing arrays can retard the footprint growth that tends to be linear with storage demand growth.
There is another argument that says that the current data center space is a sunk cost, and that reclaiming space does not directly impact cost savings. This is a valid argument if you have plenty of space in the computer room.
Sometimes, moving from DAS to SAN or centralized storage, or moving to a virtualization architecture can increase the footprint of the storage infrastructure. DAS systems have the disk local to the chassis. Virtualization adds new appliances and racks to enable this technology. Other OPEX savings may be needed to off-set growth with these types of activities. In high rent areas, space reclamation is a valid effort and can be used to offset investments in consolidation, SAN deployment and other infrastructure collapsing investments.
The following paper can be used for DC floor space or power analysis. APC Paper on Environmentals, TCO
Comments (2 )
[...] Looks like I skipped 13 and 14 since my last post, and now there are 15-17 up on Dave’s site. Sorry I know I was going to try and comment Dave, but there are so many things to discuss…. [...]
[...] Weekly Management Tasks – This one may sound very similar to item #8 the aggregate labor costs for storage management, this element is different in that we look at [...].