When it comes to the evolution of technology, or even revolutions in technology, former Time Warner Cable CTO Mike LaJoie has been in the curl of some major waves during his 25-plus years in the cable operator industry.
Among other engineering-related accomplishments, LaJoie was with Time Warner Cable when it became the first cable operator to launch a high-speed data service as well as for the rollouts of VOD, VoIP, and network DVR. But long before LaJoie retired as executive vice president and chief technology and network operations officer late last year, he started out by writing software and system code.
There are a growing number of companies in the SD-WAN space. What attracted you to Viptela?
Their solution is real, it works, and it’s effective. I ran into Viptela probably a year and half ago when they were just barely rolling out their solution. I’ve known Amir [Khan, co-founder and CEO] and Khalid [Raza, co-founder and CTO] for a while. When you run across folks that are productive technologists, you tend to keep an eye on them, and these guys are very, very bright. I was introduced to them by a mutual friend. It was just a conspiracy of things. I’m trolling around looking for stuff all of the time, and I’ve been hearing about NFV and software-defined networking for quite a while. I’ve believed in the promise of it, but I’ve always been a little bit frustrated with the manifestations that I saw. Viptela works with how they’re approaching it, so I got engaged with them.
How would SD-WAN apply in the cable operator space? Would it apply for commercial services by MSOs?
I think in its current manifestation, it’s something that would be a product that the commercial services organizations can offer to their customers. By putting a gigabit router on the edge that can be managed by an orchestration layer, as long as the device has connectivity and can phone home, it relieves you of the need for the tedious management of MPLS and all of the other stuff you have to do to maintain a virtual private network.
If you look at TCP/IP MPLS and how it works, it’s stacked in layers for a reason — so you can abstract the complexity, the layers above and below where you want to function. Often times in the past, the way we have put together a virtual private network, giving people on the edge the ability to manage a network that appears to be their own, required a lot of tight coupling of the stack, which is kind of antithetical to why the stack works.
By implementing an application orchestration stack that manages the connectivity it means that now cable can offer a set of services to larger organizations, organizations that have multiple physical sites, and have multiple network nodes working in a much simpler and more reliable way.
So as things get more complex, the key is to simplify?
Really intelligent solutions — do they make things more complicated or more simple? They should make them more simple. If they’re really smart, if they’re really the right way to do things, it should just work. That’s how it should work and that’s what Viptela’s solution does. They have thousands of sites deployed due to a simple, scalable architecture.
So further along, as the systems become more robust and as cable becomes more reliant on orchestration layers, could you actually use it as a network operator? Could you use it to implement connectivity across disparate data centers in huge terabit kinds of switching fabrics? I think the answer is yes. The reason that’s important is that the operating stack, whether its JunOS or iOS, has become so complicated in the routing fabrics. The tight coupling between the MAC layer, the physical layer, and the networking layers has made it so difficult to accommodate the growing need for bandwidth that there’s now a requirement for a solution that can plug in more fabric. It’s just going to wake up and phone home and know where it’s supposed to be. There’s promise for an operator to reduce the complexity and the cost of managing a huge network. I think that’s yet to come, but it’s coming.
You started out in software. What do you think of the move to network virtualization with software? The promise of SDN in general and how it applies to cable?
I think it will perc [percolate] in from the edge. I think the way that Viptela is approaching this, by coming up with an appliance that sits at the very, very edge of the network on customer premises, and by starting at the sweet spot where they have a gigabit kind of router, is a huge application. It will manage in a very granular way, taking very small bites. When you think about gigabit, it doesn’t sound like a small bite, but in the scope and size of all of the fabric and interconnects and the different network operations across the planet, it’s a teeny, teeny little nibble. Doing it a little nibble at a time, and having the systems perc back in from the very, edge I think is the right way to do this.
I think when you recognize the simplicity of this is as a way of setting up a network for end users that have multiple locations. Locations like large banks, distributed retail locations, government branches, schools, medical facilities, or anything else that has multiple locations, there’s a need for bandwidth and connectivity across those multiple locations. The ability to manage them as a single network, that market is huge. This will grow and become more hardened in time. I think it will perc its way back into the network and become more of a solution. You’ll no longer have to be so tightly coupled on the physical connections of the network itself in the core. The network actually works because this stuff will wake up and find itself at the edge.
I think it will take some time to perc back into an operator, but with that said, it will give the operators efficiency. It will give them the ability to offer broadband connectivity at the edge to enterprise customers. It will give enterprises the freedom to manage the networks, or have the operator manage their networks for them.
So given that it will take time, do the cable operators have an advantage working with their existing customers to implement SDN?
If the operators don’t jump in, you’ll see the enterprises do it. I think it will probably start with enterprises, as you’re starting to see with some of Viptela’s customer wins.
I’m not going to speak for what cable operators are or aren’t doing, but it’s more around the Open Networking Foundation than OpenStack, although OpenStack plays a role.
Software-defined networking and network functions virtualization, they’re not new. They’ve been gaining momentum for the last five or six years. I believe this is the beginning of a real revolution in how networks converge and how networks are built. It’s going to become much easier, much faster, and much simpler.
How are big data, the Internet of Things (IoT) and SDN/NFV related?
I don’t think those things are orthogonal. They’re not directly connected, but they are related. If you have the ability to proactively turn up and provision more and bandwidth at the edge, that certainly accommodates the Internet of Things more readily. It also supports the ability to retrieve information from test points, so all of those things kind of work together.
SDN and NFV are natural extensions of the facility that allows TCP/IP to continue to grow, and grow at such an amazing, rapid pace. It’s structured in a way that is atomic and that you can add on to in extremely small chunks relative to the big picture. It’s the same philosophy that came out of a lot of the protocols that continued to grow and blossom in the TCP/IP stack.
The modular nature and the atomic nature of what SDN and NFV really do is going to allow the network to become more dynamic again. The way it’s been growing has been phenomenal. It’s been fabulous. The expansion of JunOS and IOS [Juniper and Cisco‘s operating systems] and other stacks like them have accommodated the growth to date, but the need for more a robust, atomic solution is absolutely there.