Lesson 2: Reducing storage TCO with Storage Networks
by David Merrill on May 15, 2006
As mentioned earlier, there are 6 popular activities or investments to reduce OPEX costs in Storage. This lesson focuses on storage area networks. SANs (in the broad sense of the acronym) can reduce storage infrastructure costs over time. I do not want to imply that FC is the only type of SAN, but in the generic term a storage network could be running one or multiple transport protocols. I do not differentiate here the economic differences of NFS or iSCSI or FC; but only to say that networked storage is cheaper to own than direct connect storage.
Gartner also support this idea that DAS is cheaper to buy but more expensive to own (see IGG-09172003-01). There are many papers from vendors such as CISCO, Brocade and McData that promote the improved TCO or ROI of their product.
In summary, SANs can reduce storage infrastructure OPEX by:
Improving aggregate utilization, less stranded capacity
Improved data path availability (reduce unscheduled storage outage)
Reduce scheduled outage time for storage upgrades
Simplified management for large server pools
I wrote a paper many years ago, much of it is out-of-date, but there are still some basic principles in-force even today with SAN economics.
Comments (3 )
It seems to me that Google has figured out how to save a lot of money: get rid of storage arrays altogether. How much? Well, I estimate they’ve cut their capex in half, or more, compared to Yahoo, not to mention opex.
Read about it at http://storagemojo.com/?p=99
Do you know more deatils on how Google reduced storage OPEX, or how they compare in storage infrastructure to Yahoo? I did not see that kind of details in the article. Specifically, how did they get storage ownership so much lower?
I believe they moved away from monolithic storage (and I mean even the midrange stuff) and basically build a grid of appliances.