PICMG: A CTO's perspective
In the current environment of "open source everything," is there still a role for PICMG and open specifications? Doug Sandy, vice president of technology and CTO for PICMG, explores the past and future of PICMG within the emerging industry landscape.
I still remember my first PICMG meeting. I was only a couple years out of graduate school when my then-supervisor invited me to participate in a manufacturers’ group that would be defining a new type of industrial computer form factor. My company at the time, Pro-log, had been the inventor of the venerable STD bus; at the time, we were looking for a replacement technology that offered both interoperability with other vendors and dramatically higher performance. Although I didn’t know quite what to expect, I told my boss to count me in.
In my mind’s eye, I envisioned a room full of stodgy old engineers debating esoteric design nuances ad nauseam. What I found, however, left me happily surprised. Representatives from about ten companies faced off around rectangular tables in order to solve the real problems associated with developing a new industrial computer specification. This was a no-nonsense group, and they knew how to get the job done. By leveraging the collective expertise of each of the member companies, we were able to create the most successful industrial computing specification of the time – CompactPCI.
It was this no-nonsense approach, design expertise, and straightforward governance that led to PICMG’s success in the industry. CompactPCI grew to incorporate a suite of new features not envisioned by the original authors. Later, PICMG tackled the unique requirements of telecommunications platforms with the introduction of AdvancedTCA, AdvancedMC, and MicroTCA. Others joined the community with a commercial off-the-shelf (COTS) military focus and introduced ruggedized platforms to the PICMG portfolio. Most recently, platforms have been introduced to address transportation, computer-on-module (COM), and high-energy physics needs.
As you can imagine, a lot has changed since PICMG first opened its doors in 1994. Technologically speaking, the industry has moved from chips running at tens of megahertz to processors that clock in at over one gigahertz. Communications speeds are now in the hundreds of gigabits per second, and server storage is typically in the terabytes. With all this advancement, perhaps the biggest change has been the rise of open source initiatives. Open hardware and software projects such as Open Compute and Linux have changed the playing field for consumer and vendor alike. This age of “open” everything prompts the question: “Are specifications relevant anymore?” Perhaps this query is best answered by understanding what open source is and what it is not.
What open source is
A variety of computing-related hardware open source projects have sprung up in the past few years. In these organizations, community members collaborate to create product definitions within the scope of the project in order to serve the community and industry at large. The Open Source Hardware Association defines open source as “hardware whose design is made publicly available so that anyone can study, modify, distribute, make, and sell the design or hardware based on that design.” Requirements vary by organization, however, typical design elements of a submission include Gerber files, mechanical drawings, schematics, bills of materials, firmware and factory test code. These requirements are shown in Figure 1.
The open source hardware model works well when developers either don’t expect to reclaim their engineering expenses (e.g., hobbyists) or expect to reclaim their engineering investment in ways other than product differentiation. For these developers, open source is a vehicle to sell other values such as integrated circuits, product maintenance, installation, or manufacturing services. Making an open source submission is a method to accelerate their revenue from these sources. Alternately, large users of the technology may choose to provide their own designs to the open source community. This tactic may be used to leverage some of the design capabilities within the community and quickly commoditize the product. Of course, this method can’t be employed for products that provide real competitive advantage. Both of these strategies are a significant departure from the specifications-based approach.
The primary beneficiary of open source hardware development to date has been the hyperscale cloud computing market, where a small number of very large users seek to deploy vast quantities of very specific commodity products.
What open source is not
Open specifications, unlike open source, focus more on common interfaces than on defining finished product. Individual vendors may develop whatever they wish and remain compliant to the specification so long as the interfaces comply (Figure 2). Product interoperability between multiple vendors is a key goal of this approach. As such, open specifications favor multiple vendors that offer differentiation based on product features. Industries that require a broad ecosystem of varied components will benefit most from open specifications. Industrial automation is one such industry, as it requires many different types of I/O and processor types to serve its needs.
Military, aerospace, transportation, and the Internet of things (IoT) also pose challenges for the open-source hardware model, though for different reasons. In these markets, security and safety are key criteria. Having secure designs openly available for review and modification plays into the hands of the very elements they must be secured from. Malicious entities can use the published information to identify vulnerabilities and plan attacks. Would-be attackers can even work within the open source community in order to introduce vulnerabilities directly within the project deliverables. Open specifications avoid these issues because the security and safety critical elements remain isolated and under the control of each individual vendor.
What does the future hold?
Open source initiatives and open specifications organizations both serve the industry by promoting product ecosystems. Open source favors commoditization of specific products and ecosystems of integrators and contract manufacturers. Open specifications, on the other hand, promote broad ecosystems of highly differentiated and interoperable products delivered by equipment vendors. This difference is summarized in Table 1. Both models serve a purpose and neither model is likely to disappear any time soon.
Recognizing its role in the industry, PICMG is well positioned to continue providing open specifications to the industry in the areas that will benefit most, namely: IoT, industrial automation, aerospace, medical, transportation, physics and communications. Current activities are focused on physics, aerospace extensions to CompactPCI Serial platform, higher speed versions of MicroTCA, and improvements to the popular COM Express specification. PICMG has always been vendor-led and is always open to new projects.
Although the rectangular conference tables are gone (we now face off in conference calls), the no-nonsense, “get it done” attitude of PICMG is alive and well. With thousands of collective years of industry experience, domain-specific skills, non-onerous policies and procedures, and a focus on customer success through interoperability, PICMG is a strong choice for your next open specification project. I look forward to working with you soon.