OpenVPX Industry Working Group: Open for business, or just controversy?Story
March 17, 2009
The rhetoric is flying as intense debate rages over what might be the newest ?next-generation version of the VMEbus.?
Important editor's note: OpenSystems Media, publisher of Military Embedded Systems, has relationships with many of the companies mentioned in this column. Some of these companies are advertisers or partners, and others - such as VITA Executive Director Ray Alderman - are regular article contributors to this magazine or its sister magazine, VME and Critical Systems. It's important for you to be aware of these relationships, which might affect my "objective impartiality," even though I've done my best to represent all sides accurately and factually. -Chris A. Ciufo
We've been here before. About five years ago, VITA (formerly the VMEbus International Trade Association) announced a roadmap for the future, and its VPX (VITA 46) specification was clearly identified as the next-generation version of the venerable VME64 bus. Sure, there was some confusion over the VXS (VITA 41) specification versus VME, but the key distinction was that in order to switch your system to VPX you needed an entirely new and different backplane. VME, you might recall, is the number one open standard, COTS embedded board form factor used in harsh environment, deployed defense systems. VME is used in everything from missiles and large torpedoes, to avionics, ships, armored vehicles, and spacecraft. In a nutshell: What happens in the VME ecosystem is hugely important to the defense industry.
Programs specifying VPX include several subsystems in the Army's Future Combat Systems, the Marine Corps' G/ATOR, and the counter-IED program JCREW.
(Click graphic to zoom by 1.3x)
The not-really-open OpenVPX Industry Working Group (www.mc.com/openvpx) was recently started by Mercury Computer Systems to essentially create the system-level specification for VPX that they hope will become the next generation of VME. But wasn't that what VITA's own efforts were supposed to create? What has VITA's Standards Organization (VSO) been doing all these years? And wasn't Mercury one of the two leading proponents of VPX? (The other one is Curtiss-Wright Controls Embedded Computing or "CWCEC.") If you're confused already, it's going to get worse as I peel away the layers for you.
You see, VSO has announced about 20 different "dot specs" ranging from ANSI/VITA 46.0 "VPX: Base Specification" to the brand new VITA 46.21 (covering a distributed switch architecture). After all these years of work and what might be a million man hours by dozens of companies, why would Mercury suddenly decide to work outside of the VSO to create what looks like an entirely new next-generation VME? Is this a brilliant marketing feint to bifurcate the market? Are they so angry at the VSO that they've decided to start their own standards organization? Many VSO member companies and military primes counting on the VSO version of VPX are asking the same questions. Programs specifying VPX include several subsystems in the Army's Future Combat Systems, the Marine Corps' G/ATOR, and the counter-IED program JCREW.
Much has been written about this Mercury versus VSO situation in the past several weeks. My esteemed journalist colleagues at Military & Aerospace Electronics, John McHale and John Keller, have both weighed in on this subject. In my extensive interviews and research, I've determined that Mercury - and presumably their partners including Aitech, Boeing, GE Fanuc Intelligent Platforms, Hybricon, Tracewell, and others - sees three deficiencies with the VSO's version of VPX:
- The lack of backplane interoperability between board-level vendors
- The lack of a cohesive system architecture versus a series of board-level subsystem architectures (and the lack of data/control/management/user plane features)
- How long it's taking VSO to move from Working Group specs to Released ANSI/VITA specs while vendors need to build and ship "ratified" hardware
Citing the PCI Industrial Computer Manufacturers Group's AdvancedTCA specification set (PICMG 3.0) as the right way to leverage COTS into defense, the Industry Working Group wants to sort these issues out and present back to VSO a revised VPX system document within six months. Ironically, VITA's own management supports this unorthodox vote of "no confidence" in VSO's Standard Operating Procedure (SOP). VITA Executive Director Ray Alderman told me that while "a lot of good work" was put into characterizing the bottom-up (board-focused) VPX architecture, "that activity hasn't captured the computer science of the system itself," and the non-VSO efforts represent a "unified approach." Yikes: Sounds to me like he's saying GM spent too much time designing nuts and bolts instead of cars. Alderman asserts that PICMG's own AMC specs were created by an outside group with Intel, then later spun into PICMG and fast-tracked to become a standard.
For their part, Curtiss-Wright Controls Embedded Computing is greatly concerned, as the company is hands-down the leader in VPX and has doubtless invested millions of dollars in designing VPX boards, backplanes, and systems. Worse, Mercury's "open" Working Group won't disclose any details about their version of VPX and they won't let CWCEC join the party until sometime in March 2009 when "all will be revealed." Does this merely give Mercury and Friends time to sort out their specs without worrying about naysayers and arguments ‚Äì or time to create a system that nicely excludes CWCEC? And how can all of this possibly be condoned by VITA while leaving its own largest VPX supporter - CWCEC - out in the cold?
On the surface, everyone's motives seem pure. The OpenVPX Industry Working Group is trying to solve some legitimate technical issues that are not well defined by the existing VSO VPX dot specs. VITA is trying to capitalize on the groundswell of VPX activity by endorsing Mercury's fast-track approach. And vendors like CWCEC (and their customers) are crying foul because they're purposefully being excluded from what is supposed to be a fair and open process. I think that everyone is individually right on this one - adding up to one big controversy that you'll hear more about in weeks to come.
That was then, this is now
Clearly, ultra-recent events denote that these controversial issues must be weighing heavily on Mercury, their OpenVPX partners, and many of the end-user primes who are supporting the OpenVPX concept: The week Military Embedded Systems shipped to printer, a press conference was held to announce the opening up of the OpenVPX Industry Working Group to all VITA members in good standing. Specifically, companies such as CWCEC, TekMicro Systems, and any other interested VPX supporter can now join.
The group will hold a two-day gathering (March 16-17) preceding the next VSO meeting in Orlando on March 18-19. Following confab on the newly established OpenVPX Steering and Working Group procedures, new members will also need to sign a memorandum of understanding and possibly pay a small administrative fee to cover overhead. Since Mercury assures me that OpenVPX is trying to mirror VSO's procedures, it should be straightforward to take the group‚Äôs documentation back into VSO within six months. As well, the OpenVPX Industry Working Group isn't specifically worrying about Ex Ante and ITAR issues - but is certainly cognizant of the law since VSO will overlay its own bureaucracy onto the documents once back in VSO.
So I'm heartened that at the 11th hour, Mercury and its partners are putting the open spirit into OpenVPX by allowing any company to join. This certainly is consistent with one of their stated goals toward interoperability. In fact, Bob Ford, a Boeing representative familiar with OpenVPX, told me that interoperability between vendors will "reduce the market fragmentation in the types of [VPX] cards." Getting companies like CWCEC involved is the only way that can realistically happen.
Still, I'm troubled by the timing of all of this. Although Mercury and Friends have ultimately relented by allowing new-comers to join, the fact is that this effort was going on within Mercury from October 2008 until January 2009 when other companies quietly signed up. And another two months have passed before more companies could enter the group. Will they really have a chance to alter six months of nonpublic momentum? I suspect not, and more controversy will be brewing when the reality of forced board redesigns by non-OpenVPX vendors sinks in.
Chris A. Ciufo Group Editorial Director
Update: Though as we went to press, the OpenVPX Industry Working Group announced that all interested VITA members could join, that still doesn't resolve the controversy of those 20 dot specs in VITA and what's to become of them. Check out the April issue of VME and Critical Systems for our first Q&A with Mercury.