Aperi's Seeds of Revolution

When last I exhaled, I was ranting about the stagnant state of cross-platform storage management. You know what I am talking about: the ability to equip applications with storage provisioning and protection services that data requires over its useful life. (See Strange Fascination With Aperi.)Figure 1:

Vendors also don't see cross-platform management to be in their best interests. The Storage Management Initiative Specification (SMI-S) effort pushed by the Storage Networking Industry Association (SNIA), as well as the efforts of many platform-agnostic” management software vendors, to realize cross-platform management Nirvana were hobbled at the outset for just this reason.

Predictably, I got some pushback from management software vendors on this point. Many argued that they were quite platform-agnostic and managed storage at the logical unit number (LUN) level across every major array vendor in the market, and a lot of the smaller ones as well. I am not arguing that.

Everyone can see a LUN, of course. LUNs are outside of the hardware guy’s ability to control; they are SCSI artifacts. But viewing storage from the LUN level doesn’t get us where we need to be if we are really going to manage storage itself.LUN-based management doesn’t let us control how much of our expensive, formatted capacity is reserved by the array vendor for its own use, whether it has been allocated or not. Without this information, it is very hard to identify the real cost of gear. It doesn’t provide a reliable basis for “virtualization” (i.e., LUN aggregation). You don’t want to stripe across LUNs exposed by two different arrays having very different data and control paths, for example, lest the resulting volume prove to be unstable. And it sure as heck doesn’t give us the ability to migrate data easily or at will between two LUNs of the same capacity.

Enter Aperi. (See Aperi Appears Amid Questions.) IBM dropped it like a bombshell at Storage Networking World: an open-source initiative to create standards-based heterogeneous storage management. EMC and its paid analysts tried to dismiss it as a snub, but the idea has been kicking around for many months. IBM just gave it a voice.

The underlying acknowledgement, though this will not be found in any of the Aperi press materials, is that open sourcers might just have the right stuff when it comes to cracking all of the “cones of silence” that vendors have built around their proprietary arrays. In other words, Aperi might succeed where others have failed.

SNIA doesn’t like this notion much. My moles tell me that they are very concerned that the whole Aperi thing might cast a negative light on SMI-S, which they claim is “going just fine – despite what Toigo says.”

If “going just fine” means that the vendor community continues to sing the words out of the SNIA hymnal in public while confining their deep-seated concerns to off-the-record commentary shared privately with “hair on fire” pundits like me, then I agree. The minions are towing the line in their public comments.In fact, the situation reminds me of a visit I made to the former Soviet Union before the collapse of communism. Everyone spoke two languages – the public/party line (extolling the superiority of Soviet culture) and the private line (complaints about standing in a line for an allotment of toilet paper that ran out before the queue did).

So it is with SNIA and SMI-S. The most positive thing anyone can say is that “at least SNIA has gotten the boulder moving toward storage management standards,” as an IBM boss told me in a call today relative to Aperi. “And that is no small accomplishment in the storage industry.”

I agree. They moved a boulder. But unfortunately, like Sisyphus, the character in the Greek myth condemned to a Hell where he must push a boulder up a hill only to have it roll back down to the bottom just when he thought it had reached the summit, the SMI-S development boulder has proven quite adept at moving both ways.

Reinforcing my negative view is the commentary that has been provided to me, off the record (private speak, again), by techs and others involved in the SMI-S development process. They use expressions like:

Put another way, SMI-S might have been a noble dream, but it ignored the simple reality of the hardware vendors' preoccupation with their own top-line growth and their zero-sum mentality in the market.Can Aperi succeed where SMI-S has failed? That is the question that knowledgeable customers are asking. I have to confess, the answers provided today by IBM are less than inspiring.

I was repeatedly told by IBM that the agreement reached among the Aperi cadre comes down to three things: intent to form a community; agreement to contribute yet-to-be determined “storage information management” code to a common repository; and agreement to model the resulting activity and licensing using the Eclipse initiative as a guide.

On the first point, IBM insists that the Aperi initiative will have a “bricks and sticks” as well as a “virtual” presence. However, it remains to be decided whether the initiative will live on its own or be subsumed within SNIA as another forum. (IBM is on SNIA’s executive council, I was reminded, so discussions are ongoing with SNIA regarding the best way to proceed with Aperi.)

About the only bone they threw to those of us who have soured on SNIA, given its closed-door development methods and extremely poor outreach to the consumer, is that “it may require an [institution] outside of SNIA.” When I asked whether Aperi activities would be “conducted in the sunshine” of public scrutiny, as we are fond of saying here in Florida, they referred me to their by-laws, which were borrowed from Eclipse and “agreed to in the letter of intent” signed by all founding members. They said the organization would be “open to anyone who agrees to the by-laws.”

Here are my recommendations to Aperi, for what they're worth:First, publish your by-laws for public comment and criticism. We’ll run them for you here if you want.

Second, quit being so defensive about Aperi being characterized as an IBM venture, especially by your competitors with their own proprietary axes to grind. Big Blue is already sponsoring, and has for a long time, the open source efforts at the High Performance Computing Center at the University of New Mexico to build grid computing. Eclipse is another proof of an unselfish commitment to open standards.

You have nothing to prove to anyone and your efforts in forming Aperi are important in part because it is IBM promoting the effort.

Third, if pursuing Aperi requires that you break ranks with SNIA, feel free. No one will hold it against you.

Think about it: SNIA membership isn’t buying IBM any sort of product differentiation, and much of its work to date, even if well intentioned and occasionally visionary, has fallen prey to the proprietary goals of its ruling clique. Everyone I talk to inside and outside the industry seems to share this opinion, and among consumers SNIA is meaningless – most folks think the acronym stands for a cold remedy.Fourth, if you do decide to implement Aperi as a SNIA Forum, please remember how this strategy put to sleep the good work of the Enhanced Backup Solutions Initiative a couple years ago. EBSI, an independent initiative led by Quantum, Network Appliance, Avamar Technologies, pre-acquisition Legato, and a few others, sought simply to build reference models for use by users in comparing different data protection solutions.

Within a few weeks, there were more than 500 eager users volunteering their workloads, bandwidths, and environments to serve as testers. When SNIA saw a competitor in EBSI, they put out the word to their members to shun the initiative, denying it new vendor members and the financial resources required to fulfill its mission. Then, they twisted the arms of the companies involved with the initiative to place it under the protection of a SNIA Forum, where it has since disappeared into oblivion. Word of caution: Don’t let the same thing happen to Aperi.

Like many readers who commented on the previous column, we are temporarily suspending disbelief and hoping against hope that Aperi will succeed. The IBM folks we talked to today are still just getting their act together, and a beginning is a delicate thing when you are IBM. We will give them some time to give us some meat to go with the potatoes.

But, if this initiative becomes just another SNIF, there are those of us who are prepared to launch an open-source initiative of our own. We have already purchased a domain at www.storagerevolution.com, and we are hard at work on a requirements specification. Once that is in place, we will open the door to all storage customers and all the vendors that care to join our little revolution from below.

Bottom line: Somebody has got to do something about storage management. With 40 cents out of every dollar spent on IT hardware, having no way effectively to manage the storage stuff we are buying will eventually break all of us customers.— Jon William Toigo, Contributing Editor, Byte and Switch

Ready to storm the storage barricades? Drop Toigo a line