I have recently gotten a few emails regarding the progress (or lack thereof) in our open source effort to create a storage management framework, a process announced a few months ago and placed online at StorageRevolution.com.
One fellow went so far as to tell me that the effort was all hype because he wasn’t seeing any progress. I thought it might be interesting to look at the site and the progress that is being made.
The last several months have been spent by my team at least interviewing users of existing SRM products about what they like and don’t like about these products. What features are missing or poorly implemented. That sort of thing.
We have about 70 survey questionnaires in hand at present and I feel pretty strongly that at least 500 are needed to get down to the business of writing a consumer-driven requirements spec. I know what I think should be in the framework, but that is not the same as saying that it reflects the consumer perspective. If it does not, the entire effort is as much of a sham as SMI-S. Engineers shouldn’t seek to fix problems that no one perceives that they have. Rather, we should do a real world requirements analysis to create a picture of what is needed before we go solving imaginary problems.
This kind of outreach is hard work and it takes time. Anyone who feels that progress isn’t being made quickly enough can go screw themselves. I want to do this thing right the first time.
Code contributions have been made from folks who have workarounds to vendor obfuscation on their arrays. And a pretty good framework shell has been contributed already.
I welcome the input of readers of this blog regarding what they think is needed from a “universal” framework. Let’s get constructive.
What I am hoping we can do at the next DR/DP summit, in addition to demonstrating some low cost architectures for archive, mirroring and backup over IP infrastructure, is to demonstrate a preliminary version of our management framework. I have a lot of independent ISVs who are dying to plug in, and I have received substantial encouragement from reps of 3-letter vendors who agree (quietly, of course) that SNIA can never hope to do an open framework, given the control exerted by big vendors over the process. They are wishing me well.
I spent some time with Ken Barth, CEO of Tek-Tools, yesterday. Ken said something interesting: SMI-S is a good hook into storage when implemented with the proper goals of open management. However, many providers are very contrived and sparse in what information they offer — a function of how “closed” the vendor chooses to be with respect to the visibility of the internals of his array. This confirms what I have been saying all along.
Another big issue is that smaller consumer shops believe that management should be free, delivered with the box. No one really wants to pay for management. Most believe that a self-articulating web page (in band) is adequate, until they get to have a lot of stuff to manage. Then, they want an external management scheme that will summarize status foo.
Management is also a shifting concept. A lot of folks are looking for more than a “simple” monitor and capacity allocation solution. They are now talking a lot more about organized, policy driven data movements: copying files among disks (and removing duplicates), managing retention/deletion, archiving stale data, mirroring/backup, CAS, etc. These need to be examined for what functionality must be supported in a framework and what can be safely ignored or tabled for now.
A lot of work, folks. Not an easy task. Never thought it was.
I also heard, but have not confirmed, that Aperi has pulled up stakes at SNIA and is now moving its efforts elsewhere! Good for them!