Anyone Want a Couple of Terabytes?
by Claus Mikkelsen on November 14, 2008
I’ve spent a lot of time lately talking about a subject I’ve become quite passionate about. I wrote a white paper on this a while back, and have discussed this with many, but thought it was time to try to reach a larger audience (and kick-start my blogging activity at the same time). Those of you that may have heard me speak of canaries and storage capacity, irrational storage growth, and how our personal views and habits on storage have changed over the years, are welcome to skip to the end of this blog, but others might want to read on.
First, some “fun facts”:
• The storage industry was “invented” in 1956 when IBM announced the RAMAC 305, which quaintly held about 4-5 MB. Interestingly, it was not a lot unlike today’s enterprise arrays in terms of size and dimension.
• A 1956 system, if extrapolated to 2 TB, would require a building the size of the US Pentagon to house that capacity.
• That same system, again extrapolated to a 2 TB capacity, would have cost about $32 B USD.
• Storage aerial densities, over the last 52 years, have on average doubled every 14 months, while at the same time seen a price decline average of 30%-35% a year.
Pretty impressive facts, but why the focus on 2 TB? Well, when I originally wrote the white paper, that’s what I currently had at home. It’s up to 4 TB now since I had run out of space about 2 months ago and it was easier to buy more capacity that clean up what I already had (actually, the drives are mirrored so I now have 2 TB usable).
But the point is that I had a choice of spending many hours on a beautiful weekend cleaning up my current capacity or I could have simply gone to my local “toy store” to buy more. I chose the latter, and now stand as a horrible example for what’s wrong with storage today.
Many (most?) of our customers are doing the same. With a 30%-40% utilization rate, we continue to spend on additional capacity. Rather than spending time cleaning up what we have, we continue to grow. What has happened (and this is when the canary died), is that capacity is increasingly being viewed as unlimited and the cost irrelevant. Certainly we don’t want to inhibit business growth (or ruin a nice weekend) by failing to add capacity. And as we continue to add capacity and data, isn’t it becoming increasingly difficult to use what we have, as in find that important email or Powerpoint slide?
This trend might point to chaos if nothing is done, and in fact it could. But the storage industry (hardware and software, collectively) is leading the way. New (and some less new) technologies, if adopted can in fact start mitigating the effects of this irrational growth. This is an area where HDS and Hitachi are clearly leading the way. Storage virtualization has shown to ease management costs, reduce complexity, improve utilization rates, and improve TCO. Dynamic provisioning (or thin provisioning) can not only dramatically improve utilization rates as well, but provide some serious performance improvements to cache-unfriendly workloads, and again overall improving TCO and ROI. And if you believe my boss Hu (and I do), improve ROA (return on assets). De-duplication technology can sharply reduce the consumption of capacity for backup and archiving tasks. Data discovery technology finally gives a chance to actually use the many TB or PB that we’ve archived. And finally, wrapped around all of these technologies, and more, we have Services Oriented Storage Solutions (SOSS) to start driving the knowledge and requirements to the application level. In the end, we don’t want you to have to concern yourself with storage; we’ll take on that task. You should concern yourself with the applications that drive your business.
And finally, if only these technologies were available to us consumers, I could have save a few hundred dollars and a trip to the “toy store” a few weeks ago.



