Arpit Joshipura became the Linux Foundation’s new general manager for networking and orchestration in December 2016. He’s tasked with a pretty tall order. He needs to harmonize all the different Linux Foundation open source groups that are working on aspects of network virtualization.
Joshipura may be the right person for the job as his 30 years of experience is broad — ranging from engineering, to management, to chief marketing officer (CMO) roles. Most recently he was VP of marketing with Prevoty, an application security company. Prior to that he served as VP of marketing at Dell after the company acquired Force10 Networks, where he had been CMO.
But he also boasts a technical pedigree, with an advanced degree in computer science and engineering experience at Nortel and a couple of start-ups.
Is your role at the Linux Foundation a new position?
Joshipura: That is correct. This has been created late last year, and the role of this position is essentially to look at where open source networking has come from and where it needs to be taken. This includes all the components of open source networking up and down the stack: from data plane, to control plane, to orchestration, to management and architecture, end-to-end.
So, where does open source networking need to be taken?
Joshipura: Phase one of open source networking was all around: hey, let’s figure out a different way than asking the classic five to 10 vendors to build stuff. Let’s figure out an open source way to do it. We’re past the first wave. And now we’re past the second wave, which is: are these components production ready? Each of the individual components is production ready. What we need to do in the third wave is to really harmonize the components into a more end-to-end solution. That’s the key requirement I’m passionate about. I want to make sure that the end user has the ability to get a production-ready solution from open source networking.
That sounds difficult. How do you oversee that harmonization?
Joshipura: The method by which you can do that is driven from the requirements of the end user. If you take an AT&T or a Verizon, you have to look at it from their service perspective. What products do they sell? Verizon will not come in and say, ‘Hey, do you want to buy an ODL controller?’ They’ll say, ‘Can I give you network-on-demand?’ From that perspective, they sell services. You have to map these services to specific use cases and the components you need to make that service come alive. The creation of this position is because I now have influence and relationships with all these networking projects. We are going to bring them together at a leadership level with a cohesive architecture for each of these use cases, with a framework that dictates how they interoperate between each other.
Do you consider the Linux Foundation to be an umbrella organization over a lot of the open source SDN/NFV work that’s being done?
Joshipura: Yes. One of my compelling reasons to join is because it has an established image and name. There is a lot of value-add in terms of governance, events, and marketing. And just the fact that we can bring the right people to the table to solve a problem. After this phase, we will see the hockey stick of mass adoption.
What about MEF? It looks kind of like an umbrella organization, too, with its Lifecycle Services Orchestration (LSO)?
Joshipura: It’s complementary. There are really three parties that are aiming to solve the network and orchestration challenge. There’s open source, there’s standards, and then there are vendors. The three are not mutually exclusive. They’re not always complementary, nor always competitive. At the Linux Foundation, we do open source. What MEF does is standards. They have historically done standards. They define the specification, and they define the interfaces. The layer at which MEF touches and is complementary to us is at the Open-O level. And OPNFV pulls OPEN-O in. There are five to 10 standards in networking that matter. The whole LSO project is connecting the users into the infrastructure. That’s where Open-O fits in.
How do all these projects at the Linux Foundation fit together?
Joshipura: One of my goals is to simplify how the projects fit together because it’s not clear to the outside world. This picture gives you a sense of what these projects are doing. (Click here for the graphic on the Linux Foundation website).
At the data plane services level, now that you have commercial silicon and white box servers, you want to do data plane acceleration. You want to make it run really fast like it used to in a proprietary world. These projects — IO Visor and FD.io — are aimed at solving those problems. Each of them solves similar, but different problems, but they are in the same zone.
The next level up is the controller projects. Three years ago SDN came in, and it detached the data plane from the control plane. The control plane was driven by organizations like ONOS, ODL [the OpenDaylight Project], and other controllers like OpenContrail from Juniper, for example. That layer has also gone through a set of maturity. ODL is now at a production-ready level with its latest release, SR4. It’s got the most contributors or committers than any open source project, one of the very successful projects.
If we move up the stack there’s a couple of layers missing: that’s the operating system. You have an OS for the networking devices, which in a Linux Foundation context is a project called OpenSwitch. It’s an OS that basically powers the switches.
Then on top of that for carriers you have a framework called OPNFV. That’s a process set up where a carrier can pull all the different pieces together to make it into a product. This is one of the very successful carrier-driven open source projects.
Now the problem has shifted and moved up the stack into orchestration. We have projects called Open-O and ECOMP from AT&T that they have announced they are moving to the Linux Foundation. This orchestration, management, and policy software is a complex area. I believe 2017 is going to be the year where we solve those challenges at this orchestration and management and policy level. This is at the heart of where infrastructure meets the end user — their OSSs and BSSs. A lot of good momentum and support is coming from carriers.
If you look at the stack on top of it, this is how we take the rest of the pieces in networking and program an enterprise application, for example, or move it to the cloud or containers. There are adjacent projects in networking that make up the whole stack. And then there’s security, operations, training, and governance.
So just to summarize your main focus at the Linux Foundation?
Joshipura: We want production-ready architectures and product-ready solutions and to get that momentum up for open source adoption. We’re in the third phase of open source networking, from incubation, to individual components being ready, to the collective solution being ready. That’s what this phase is about. It’s all about scale.