Inside RDK

Inside RDK – I’m in RDK: Jim Crammond, Intel

Jim Crammond, is, in a sense, an “RDK Lifer,” having been involved with it since the very beginning. It was a natural progression, given his trajectory through the field of advanced set-tops — starting with is work on the back-end and client/box elements of the then-groundbreaking Moxi Media Center. The “Moxi Box” ultimately became the fulcrum for an industry-wide shift to faster, more powerful processors, to give operators more competitive clout. His last project with Digeo (which acquired Moxi in 2002) was to develop an IP client box — which, in 2006, was unheard of. His experience in working “down to the metal” is what got Intel’s attention — he joined the semiconductor giant in 2007, and has led its RDK directives ever since. For those reasons, and more, he’s our featured “I’m in RDK” exec for this edition.

Q: When did you know you were a technology kind of guy?

Crammond: My older brother was the one who built radios, stuff like that. I wanted to be a cartographer. Then, somehow, I was persuaded to not go in that direction, and ended up studying computer science.

Q: Connect the dots, from Moxi/Digeo to Intel and RDK?

Crammond: Sure. At Moxi, I built the back end systems for services like the EPG (Electronic Program Guide.) When Digeo bought Moxi, I moved on to doing the work on the client side — from the actual set-top box, to VOD integration, to running the team that did all the middleware work, down to the metal. When Digeo closed their Palo Alto office, I didn’t want to move up to Washington — if I wanted a lot of rain, I’d go back to England! That’s when I moved to Intel.

Q. “Down to the metal,” sounds like a fitting segue to the RDK!

Crammond: Yes. It traces back to the 2009 timeframe. We had a chip called Canmore — an interesting chip, because it was way more powerful than any set-top chips at the time. We engaged with Vividlogic (now a part of SeaChange) to port the cable stack onto the platform. We took that to Comcast, which had started to a project called “Parker.” That effort ultimately used the Canmore chip and Vividlogic as the base stack. That became RDK.

Q. Were there parallel thought processes happening in the marketplace?

Crammond: Yes. A couple things happened. From an Intel perspective, and at the time, it seemed like everyone wanted a different stack — the operators, the OEMs. Tons of fragmentation. Each stack was a multimillion-dollar investment. It was very difficult for a new entrant to really break in.

When we explained this problem to Comcast, they were recognizing that they had the same issue — all these different stacks that their suppliers were bringing in. They’d spent a lot of time trying to make things work across different platforms. At some point, we both recognized that RDK would be a way to potentially reduce the fragmentation and help everyone go faster.

Q. To what extent is RDK a part of your day-to-day?

Crammond: It feels like it’s every day! I mean, I think it’s the focus, or re-focus, on broadband. A year ago, it wasn’t clear that the industry was going to coalesce behind RDK-B. For the Intel Puma 7, we knew one of our customers would require RDK-B but we also had our own stack for Puma 6 so the question was whether we had to support both for the new generation of chip. We made it clear that we want to make RDK-B our main stack. Eventually, there were enough people and companies saying either “I want RDK-B,” or “I’ll stick with my OEM,” which was enough for us to proceed.

Q. How do you decide what all goes on-chip? Besides the primary RDK-B stack, for instance?

Crammond: That’s the art of it. We come up with what we call “landing zones.” We talk to the operators about them. The challenge is, operators have a good view of what they want to do in the next 12-18 months, but after that, it can get pretty fuzzy. Some are better, but for the most part, operators have a two-year window of knowing what they want to do. Our chip development cycles are much longer than that. We have to project what we think they’ll want. Some of it is fairly obvious, some of it is less so. The art is predicting what things are going to be needed two plus years from now, and how.

Q. What would you highlight as an industry advance that’s directly attributable to the RDK?

Crammond: I think it’s the lack of fragmentation. On the video side, it’s more clear – the de-fragmentation of the stack. It’s something we’ll see on the broadband side, to some extent, as well. That’s a big benefit. Another big benefit is that it allows for multiple parties to be developing in the same environment. The RDK helps to promote collaboration.

Q. What’s something about you that not many people know?

Crammond: Part of my role at Intel, in the beginning, was to educate the company about how the cable industry works. Recently, I was given a map of my node, where I live — you can drill down and see the amplifiers, all the elements of the system. It’s very handy, showing pictures of things hanging on poles, and what their role is.

Q. What are you doing when you’re not working?

Crammond: I live in Palo Alto. We have three kids — my daughter just graduated from Glasgow University, in Scotland, with a Bachelor’s Degree in mathematics. We have twin boys, who will be 20 soon — one’s in college in Chico (Calif.), and the other in Embry-Riddle, Arizona. When I’m not working, I’m big into skiing. It’s one of my main winter activities. And just the outdoors in general — hiking, this year I have been doing sailing ASA certifications. My wife, Lindy — her family has a boat, in Scotland. We go almost every year. Part of learning to sail is learning about knots. It used to be that I’d tie some kind of knot, and someone from the family would come up later and re-tie it. I’m better at them now.

Q. What’s something about you that not many people know?

Crammond: A long time ago, I was most known as a “sendmail guru” – referring to the server for delivering email, which first appeared in Unix in the early ‘80s, and is still used in many Linux systems today. In the ‘80s we had a particular issue in the UK —our domain names were backwards (e.g. uk.ac.ic.doc rather than doc.ic.ac.uk). I developed a “UK-sendmail” package that was widely adopted at the time and could handle this, along with things like uucp address syntax, bitnet and so on. When I left Imperial College I sent a funny farewell “mail administrator’s memo” to the mailing list I maintained. Someone posted it on the rec.humor newsgroup and you can still find it on google today.