Imagine, if you will, a profitable and growing industry. Since it is profitable and growing, the key industry players keep investing, producing new variants to keep buyers interested. Buyers are interested, so sales keep rising. And then one day demand flattens and then declines. Vendors are at first complacent, then worried, then panicked. Promoting new price cuts, customers still stay away. Ten years later the boom years are a distant memory.
Mainframes, minicomputers, ERP software or breakfast of champions?
Try the US ready-to-eat breakfast cereal industry. Essentially, after 15 years of price increases averaging 7% a year, people decided they didn’t need to eat breakfast cereal and stopped. The decline continues to this day.
This is the invisible hand of the market at work. Good times can turn to bad times very quickly. I think that the storage array industry is on the brink of just such a cliff. This isn’t like the dot-bomb era of 2001-2002, where tons of brand new and nearly new storage equipment got dumped on the market by liquidating startups. I believe there is going to be a fundamental realignment in the storage array business, similar to the transition from direct attach disks to direct attach arrays. This transition will be faster though because, unlike past storage evolutions, the next generation will have the signal advantage of lower cost.
Signs and portents
We’ve seen major IT transitions before, from mainframes to minicomputers, from minis to PCs, from hierarchical databases to relational DBMS’s. Lots of transitions, and the signs are there that another is in the works. What signs?
- Major customers are rolling their own storage without arrays: Google, Amazon, Cnet and dozens more. There are new paradigms that storage intensive applications are using. The big concern: none of the big storage companies are leading or, AFAIK, even participating, with the possible exception of Sun. This is the storage equivalent of PC companies eating the lunch of the major computer companies.
- New vendors, like Rackable and Verari, are offering low-cost storage options that take advantage of new paradigms, and more are in the pipeline
- CIOs are recognizing that spending top dollar to store little used data doesn’t make sense. Industry proclamations on “tiered storage” and “ILM” aren’t backed up by the tools to make them work.
Storage clusters are getting respectable
With NetApp offering them and Isilon’s IPO they can’t be dismissed as a passing fad. As costs decline, more and more CIOs will find it is cheaper to store information than to (IL)Manage it. And the self-managing capabilities of clusters make them even more attractive.
Comments welcome, of course.