Original URL: http://www.theregister.co.uk/2009/01/30/aperi_is_dead/

Aperi dies on its arse

Time-wasting diversion keels over

By Chris Mellor

Posted in Storage, 30th January 2009 10:02 GMT

Like Monty Python's famous dead parrot, the futile Aperi open source storage system management project has fluttered to earth because IBM has removed the funding nail that was keeping it upright. Aperi is now openly dead for all the world to see.

The whole idea has been a waste of time and effort from start to finish. IBM led a group including Brocade, CA, Cisco, Emulex, Fujitsu, LSI Logic, NetApp and Sun, which set up Aperi in October 2005. The setting up effectively rejected HP and EMC from the group. HP had bought the SMI-S-compliant AppIQ storage management company earlier in September that year, SMI-S being the SNIA's (Storage Networking Industry Association) storage system management standard initiative, widely held then as now to be inneffective.

It was suspected that IBM set up the Aperi group as a response to this.

NetApp's CTO, Jay Kidd, has separated storage system management into view APIs and control APIs. A view API is an interface for a system management product to look at an EMC, an HP or other storage product and find out what is going on. A control API is the means whereby it can tell the storage product to do something and its these control APIs that tend to be regarded as proprietary property and kept close to suppliers' chests. SMI-S has ineffective control API support from vendors.

As an example of this, read storage practitioner Chris Evans' blog entry on the topic.

Aperi was set up to remedy some SMI-S ills and also, fairly obviously, as a competitive marketing weapon to beat EMC and HP with. However, as IBM knew - it's blindingly obvious - you can't have a storage standard if lots of the main players don't support it, and if the ones that do aren't serious about funding it.

So Aperi set out on its blundering path to the storage cemetery. Symantec refused to join in, saying Aperi wasn't properly aligned to the SNIA's SMI-S product. The SNIA, of course, being an industry body, couldn't tell half its members to stop an effort that was going to divide the storage management area and be a self-defeating effort. It was hard enough getting SMI-S off the ground without these other SNIA members trying to fly their own alternate kite.

In June 2006 Sun abruptly left Aperi and offered its support to the primacy of the SNIA and its SMI-S project. IBM offloaded Aperi onto the Eclipse Foundation and said Aperi would have a formal relationship with the SNIA. Novell also joined the Aperi 'community' then.

IBM said at the time: "This is the latest step in Aperi's efforts to give customers more choices for deploying open storage infrastructure software - based on an industry-standard platform developed by the open source community... IBM plans to contribute more than one million lines of code from its TotalStorage Productivity Center software to the proposed Eclipse project." Wow.

As for the SNIA, IBM said: "While SMI-S is the open standard specification that SNIA members support and drive, Aperi is the open source implementation of that standard. By providing a tested implementation of SMI-S, which standardizes storage management technologies for storage hardware interfaces, Aperi will drive greater industry support and wider adoption of SMI-S." Noble words - but they turned out to be meaningless.

June 2007 came around and the initial Aperi code was ready to download for testing a development. But nobody wanted it.

The 'community' concept here was a very one way thing. It turns out that IBM was the only committed funder of the Aperi project in Eclipse, as an email to the Aperi-news distribution list in the Eclipse organisation makes clear.

Its sender Allen Heitman, with an email address ending in @us.ibm.com, said:

IBM has decided not to staff the Aperi project for 2009. ... Currently, the Aperi project has been an Incubation project under the Eclipse Technology Project for several years. Incubation is a phase and not a place. Since IBM is the only company providing active committers for over the last year, I propose that we move the project from Incubation phase to the Archived state.

The original goal of Aperi was to build a diverse community of committers to help share the burden of the "framework" in order to free up resources in all member companies to expand on the higher level functionality. Over the years, we have not been able to build up this diverse community despite many efforts. The interest just wasn't there.

The interest just wasn't there? Many people could have told IBM that. And what of the other proud Aperi group members? What resources did they put in? Not a lot if "the interest just wasn't there".

It's a great shame. The goal of having a multi-vendor, standards-based system management framework to enable management of heterogenous hardware and software vendor storage products is one that customers really, really do want. But the storage industry, collectively, really, really does not want to supply it. Generally, suppliers don't want to open up their products to the threat of replacement by competing products. Nor do they want to lose proprietary storage system management product revenue to an open source product alternative. They're not charities.

In a way, Aperi was an admirable project at first and, if a groundswell of open source support for the thing had risen, then vendors outside Aperi could have felt the need both to join in and to help the SNIA make SMI-S a real and useful standard instead of the toothless and fairly mangy hound it is now. The SNIA will protest and say it isn't toothless and it isn't mangy, but practitioners like Chris Evans and others will disagree and point to chapter and verse where it just does not suffice. This is a dog that does not bark.

The SNIA can't force its members to pony up and cut the SMI-S cloth to suit storage customers; it is a trade body after all, not a police force. Aperi was an attempt that was doomed to failure from the start, that wasted resources, blundered around and then headed unerringly towards storage hell.

It wasted the time of committed and hard-working people who supported it but who were, it may be judged, just tools in a marketing battle between IBM on the one hand, and EMC and HP on the other. That is a pity. ®