In-Depth

Oops—They’re Doing It Again

First, SMI-S, then Continuous Data Protection, and now ILM: SNIA’s on another meaningless standards-making jag

• By Jon William Toigo
• 07/19/2005

My previous missives on information lifecycle management (ILM) and data management have prompted many e-mails from readers. The latest, from a large and well-known financial services company, summed up the problem. The writer started by quoting my previous piece, “Data inherits its characteristics, like so much DNA, from the business processes and applications that generate it. From a storage perspective, data is just a bunch of ones and zeros. In the final analysis, storage management software has little or nothing to do with data management.”

He then added some thoughts of his own. “This is right on and I have seen first hand the impact of the vendor approach to solving these problems. ILM for us is a journey with no destination in site [sic]. This is probably the best and only way for us to approach the implementation until there are broader standards and smarter (business process focused) software.”

Another reader copied me on a PDF file: The Storage Networking Industry Association (SNIA) ILM Roadmap, dated 12 October 2004. You can download and read it for yourself at http://www.snia.org/tech_activities/dmf/docs/SNIA-DMF_ILM-Roadmap_20051013.pdf. He included some caustic remarks I won’t repeat here, but I had a look for myself—and you should, too.

The document provides a “definition” of ILM that sounds like a verbose description designed to be inclusive of the broadest number of members' products possible. (Remember, SNIA is not a standards group or consumer advocate, but an industry association.)

Things go rapidly downhill from there. For one, the document sets forth an ILM model that overly complicates the business analysis side of the ILM effort, while dumbing down the all-important instrumentality for making data management real at the technology level. It divides the business side of ILM into a “business framework” with lots of lines running between a box called “business processes,” one called “business objectives,” and a third called “Goals Management.” No explanation is provided for what these objects and flows mean, but the diagram suggests that SNIA is trying very hard to look like they can rise above storage infrastructure and focus on business needs for a change. This is right-headed thinking, since a business-process focus is an absolute requirement if data management, ILM, and related efforts are to produce workable strategies.

The lower half of the model is called “ILM framework” and consists of blocks labeled "applications," "information management services," "data management services," and "IT infrastructure" (broken down as network, compute, and storage infrastructure). No clue as to what the “information management services” block means, and a real question exists why there are separate storage and networking blocks. Aren’t they supposed to be merged in “storage networking” environments?

The real purpose of the paper is in the next couple of pages, which speak to yet another block diagram. Here is where the entire ILM thing falls apart and SNIA reveals its true colors.

There's a description of a multi-phase ILM vision that begins with the deployment of storage technology and its squaring away in terms of SRM (or is it SMI-S?). This effort alone, SNIA’s writers concede, will take a very long time because heterogenous management in storage infrastructure remains elusive. (They stop short, of course, of pointing the finger at the vendor membership of SNIA, who resist cross-platform management of their wares.)

Once phase one is accomplished, whenever that is, we are ready to build on this solid and stable storage infrastructure, the document continues. The vendors add value with standards and service level objectives (SLOs). We map data to SLOs through an undefined process based on intrinsic (how much money will the company lose if this data is unavailable) and extrinsic (risk to company image if the data is lost) classification criteria. It seems to me that readers of this column could suggest a far more granular set of criteria for data classification, but alas, we are not part of SNIA’s Data Management Forum.

SNIA says that they are going to eventually come up with a data classification scheme for us so we don’t have to spend so much money on consultants. I suspect that this will mean that agreements will be reached between vendors to decide in advance who gets to host what class of data: a great money pump for the vendor if SNIA succeeds in wrapping this stuff in the flag of “standards.” By the way, SLO is a good acronym because getting to this point will likely be a SLOW process.

Once all of the classifying is done, SNIA says we are going to start building ILM by “homogeneous platform stack”—or words to that effect. Once SNIA completes its valuable work, you will be able to organize and manage data across your EMC platform, and your HDS platform, and your IBM platform, and each of your other platforms individually and separately—as a set of stovepipe arrangements.

Quizzically, the author’s explanation for this arrangement is that experience has shown this to be a much easier approach. He stops short of stating that anything is easier than trying to get myopic and self-serving vendors to work and play well together. What he ignores is that such behavior is pretty much how data management is being done today-- where it is being done at all.

I will not dwell on the last two phases, which involve SNIA gifting us with yet-to-be-developed ILM tools and figuring out some way to get all of the stovepipe arrangements interoperating efficiently. The language here is as vague as anything you hear at a pitch meeting in a venture capitalist’s office: a scenario that is obviously very familiar to the author of the document.

SNIA invites what I assume are plain folks like us to participate in the Data Management Forum, though I am not sure what they have in mind: more “participants” paying membership dues or non-paying participants who will be afforded a peanut-gallery view of the process. Most people I know have better things to do, such as figuring out for themselves how they will manage data to business process requirements in their own shops.

Bottom line: SNIA’s bottom-up approach to ILM is a vendor-driven silliness that ignores the obvious business-focused dynamic of real data management. ILM starts with business process deconstruction and data analysis, not with storage stovepiping. Your feedback is welcome: jtoigo@toigopartners.com.