I’ve been catching up on a few RSS feeds today (I’m currently standing at around 8000 unread articles) and was browsing through this from Barry Burke (TSA) -1.051: skate to where the puck is going. Whilst the content of the article itself isn’t that illuminating (a discussion on how wonderful EMC are to be selling bigger SSDs in their arrays), what caught my eye was the mention of the term “slot cost”.
For years I’ve been banging on about slot cost. Imagine a storage array that supports 1000 drives. The cost of the chassis, front-end ports, power/cooling etc all contribute to the overall running of an array. Selecting the components that stay constant, irrespective of the number or size of disks in the array, the effective “overhead” of maintaining those disks can be calculated. That overhead is there whether there’s 1 or 1000 disks in the array and is constant whether there are 73GB or 1TB drives in use. The “slot cost” is the cost of servicing that slot (or drive) in an array.
I like to use slot cost (or the relative efficiency of a slot) as a measure of how efficiently configured my storage array is. Arrays with a higher relative slot cost stand out as candidates for replacement or upgrade. Here are some example metrics:
- GB/slot (bigger is better)
- KVA/slot (smaller is better)
- £/slot (smaller is better)
These calculations can be made against both available slots and slots in use.
Another interesting idea is to measure the slot cost against different storage vendors and technologies – compare DMX/USP/DS8000 for example or modular versus monolithic. This gives a good comparison of the overall cost of maintaining that storage and can highlight where data could be more appropriately stored on a cheaper platform.
-
Pete Steege
-
Dale Clutterbuck
-
Chris Evans
-
Chris Evans
