Carter: OCP Seeks to Broaden Reach at OpenStack

BOSTON -- Bill Carter, CTO of the Open Compute Project, carried the open hardware flag at a major open software conference last month.

Along with several network carriers, plenty of open source organizations were out in full force at the OpenStack Summit in May to celebrate the concepts of open source software. Meanwhile, Facebook's Open Compute Project (OCP), which instead focuses on open hardware solutions and projects, maintained an exhibitor's booth in an out-of-the-way yet still crowded side aisle of the expo floor.

Because of his unique perspective, Telco Transformation took time at the Summit to sit down with Carter. In part one of this Q&A, Carter discusses the Open Compute Project's purpose, philosophy, and partnership with network carriers as well as their partners' pain points. In part two of this Q&A, Carter delves into the economics of vendor lock-in and open solutions.

Telco Transformation: What is your aim here this week at OpenStack?

Bill Carter: A couple of things. We are continually trying to broaden our community, and the OpenStack community is supportive of open approaches. I think we have a lot of common interests with this community. And it's a complementary community. We're focused on open hardware; they're focused on an open software stack.

TT: Working with network carriers, what is the importance, from your perspective, of these open approaches?

BC: Well, the network carriers expressed interest in evolving and transforming their operations -- both their infrastructure operations and their data center offerings. That transformation started several years ago, and they approached the Open Compute Project to get more information. And we decided that we would actually put together a user community focused on the carriers. We call that our Telco Project. (See Telcos Board OCP Bandwagon.) And it grew from just a few dozen people three years ago to a mail-list membership of probably approaching 400 people today. We have engineering workshops once or twice a year that are specifically for that project -- to allow them to get together and share their business goals around open hardware and open software, and allow them to collaborate on meeting those objectives.

TT: In that space, what are the pain points that have repeatedly come up? Whether they're opportunities, whether they're challenges, what are the hot button topics you've been seeing among your members or partners?

BC: The carriers have expressed the desire to have more choices for hardware, and specifically a desire to have an open network solution -- also referred to as a disaggregated solution -- that scales from their data center operations throughout their infrastructure.

TT: Can you tell me what you mean, in this context, by "disaggregated?"

BC: Yeah, disaggregated is a term that Facebook started using several years ago when they put the [Open Compute Project] together. And it's a term that we refer to where we're taking the hardware and the software and we're splitting that up. So you can independently choose the best hardware to run, and independently choose the best firmware and software stack to run on that hardware. We've also taken it in a few other steps where we're segregating compute, storage, and networking functions within the rack. If you think of the traditional server, we've taken the storage out of the server, and we now have mass storage. And that's another way to disaggregate the hardware.

TT: So to speak of this disaggregation and this demand for more choices for hardware, where do you stand on the vendor lock-in issue?

BC: Well, if I use the network disaggregation as an example, when we decouple the hardware and software, we allow the customer -- in this case, the carriers -- to choose the best hardware platform. And we now have dozens of choices there that deliver a variety of I/O configurations. And they're able to choose the best piece of hardware for their application, and that hardware could be common throughout the data center and extend into their infrastructure and toward the edge of the network -- or it could be different. At the same time, we allow them to now choose the best software solution. In some cases, that software solution could be a proprietary solution that's downloaded and installed onto it. It could also be an open solution. And then there's another option where they use an open sourced solution, but it comes bundled with support and a service offering.

So the ultimate goal there is that they are trying to avoid vendor lock-in. Now we allow them to choose the best hardware through this community, and we allow them to choose the best software stack to run on that -- and it doesn't necessarily preclude proprietary solutions -- because there are solutions, proprietary solutions, that can run on open hardware.

— Joe Stanganelli, Contributing Writer, Telco Transformation