Sizing Up COTS: an SA Forum Perspective

I recently spent a couple of hours with the CTO of a Tier 1 telecom equipment manufacturer (TEM) discussing the state of the commercial-off-the-shelf (COTS) industry.  Going into the meeting I knew that he was concerned about his R&D cost structure – especially platform development – and was actively looking at acquiring as many of the components as he could from the industry and stop spending time and resources building them in-house. He had also told me earlier that he wished to create a common platform for the various applications and services his company developed and sold to service providers.
“We have had 6 different groups with a total of over 400 engineers developing platforms using different hardware, different operating systems, and 6 different middleware implementations,” he said. He went on to say that he was committed to reducing this duplication by creating a common application platform that various groups within the company could use. Moreover, he said he was open to using as many COTS building blocks as made sense.

It made a lot of sense to me, and I knew I had a motivated listener. So I spoke to him about what a great job PICMG had done defining the requirements of xTCA systems, and subsequently catalyzing a whole industry where there are now several significant players providing rock-solid ATCA gear including chassis, blades, etc.  And I shared with him the bullish projections about the uptake of these systems that I have been hearing from my friends at Light Reading.  Then I shared with him my own personal experience with how Linux, and especially Carrier Grade Linux (CGL), have so quickly become a requirement for many telecom applications within a few short years. Of course he knew all that – but he listened with interest.

I went on and mentioned to him that he could not only acquire a standards-based hardware platform running any flavor of Linux from one of many distributors, but that he could also acquire a sophisticated set of middleware capabilities based on the Service Availability Forum specifications so that his team no longer had to develop these capabilities in-house. All of this seemed to resonate with him.

“And you can get a vertically-integrated and pre-tested application-ready platform from your COTS ecosystem – today! All based on open standards,” I added. Then I listed the various system suppliers that I knew of that are offering pre-integrated vertical stacks – the hardware, the operating system, and the middleware services – all based on the available COTS building blocks.
We were sitting in his spacious and elegantly appointed 20th floor office, and the evening hour was fast approaching. After a pause, a penetrating look at me, and then gazing through his office window, he said thoughtfully, “You know Asif, this is all very intriguing. I cannot help but think, however, that this is like happiness. We all want to be happy, but not many of us believe we can buy it.”

It took another lively half an hour before I clearly understood what he meant. He said he was convinced, and now predisposed, to acquiring an application-ready platform as opposed to building it provided it met their requirements. He said that the many presentations he had sat through on this topic had not fully addressed some of his key concerns. Here is my take away from what he said:

• Compliance with industry standards is important and necessary but not sufficient. Various layers in the stack must first meet the key functional and non-functional requirements.
• Through years of investment, TEMs have developed and enhanced key software components – especially middleware – and they would like to maintain some of these components moving forward. Often they are a part of their competitive differentiation, so the COTS-based application-ready platform must allow for that.
• Significant effort may be involved when transitioning an existing application to the COTS-based platform. It is important to know specifically the cost benefit of such a transition prior to launching such a project.
• And finally, it is important to clearly understand if this transition can be made without introducing de-stabilization, cost overrun, and schedule delays.

As I considered his points, I realized that he was absolutely right.  I believe those of us who consider ourselves part of the COTS ecosystems we are promoting, can and should take the above challenges head-on. At the end, we may not be able to sell happiness, but an application-ready platform is a definite possibility.