Ciena's CTO Steven Alexander talked with Silverlinings about the emergence of the edge cloud, how cloud just "needs to move closer to the customer," the importance of low latency at the edge and the key services enterprise customers are asking their service providers for.
Silverlinings sat down with Alexander at the vendor's Vectors 2023 customer event in Ottawa, Canada, in early June, after a full day of touring the company's R&D facilities. He gave us the run-down on what it means to be CTO for the company, insight into his "mean-time-to-cloud" approach when it comes to network architecture development and his views on what he calls the "yours, mine and theirs model" for services.
Listen to the interview above or read the transcript below. You can also listen to our interview with the company’s CMO Rebecca Smith in a separate episode.
Don’t miss an episode of Cloud 9. Subscribe to our YouTube channel, and for daily content, visit Silverlinings, and follow us on LinkedIn, Twitter.
Liz Coyne: Hi. This is Liz Coyne with Silverlinings and I'm here with Steve Alexander, CTO, Ciena, at the Ottawa campus for Vectors 2023. Steve, thanks for joining me today.
Steve Alexander: It's pleasure to be here — live from Vectors!
Coyne: Live from Vectors! So, tell me a little bit about yourself.
Alexander: Oh, let's see. I've been at the company a while, since basically about 1994. I came in with the first group of engineers. Our first products out the door were around wavelength division multiplexing [WDM], and I've had a variety of roles ever since. I like to describe it as certainly technology, which is my first love in this in this space, but also, I'm traveling, talking, trouble-making, troubleshooting and the occasional tantrum.
Coyne: So we were preparing for recording here, and you started giving me some insight into what you're talking about here at Vectors with your customers, and rather than specific technology and the product part, you're really big picture, obviously as CTO. So tell me a little bit about what the biggest big picture concerns are from your customers. What are they seeing as the biggest challenges coming down the pike?
Alexander: Well, it varies a little bit, of course, depending upon where they are and their own issues, but they always typically come down into a conversation around capacity — the concept of the network needing to just get faster.
Another thing that they like to talk about is what happens with the cloud because the cloud in many ways needs to come closer to them, right? That’s kind of an understanding that latency now is an important parameter that they need to understand and measure, and in many cases, plan for and get some other instances even built differently in order to accommodate.
There are also general discussions around what types of services are going to be available in the future. What do we see from their peers? What do we see happening from other customers that we talked to, and what the evolution of business services is going to look like? And so, it's been a great set of conversations with folks because one of the things that's really nice about Vectors is talking directly to the development engineers. Customers can come in here and have a dialogue directly with the people that make technology choices. They can ask, ‘Well, why did you do this? Why didn't you do that? How do you consider the following?’ And it's a two-way street. It's great for our engineers to hear directly from the customers what problems are they facing. You know, engineers need to kind of exist to help solve problems right so understanding what the problems are is critical.
Coyne: Coming off a pandemic, engineers must have been existing in their own little bubble. This is the first time you've been able to talk to your customers in a number of years.
Alexander: That's right. That's why I’m the chief traveling officer, and it’s great. I think it's four years since we've been able to have the labs open like this and bring people in. And it was really fascinating. We've been able to develop great technologies for increasing capacity of the fiber networks, Ciena’s major contribution was WDM, back in the middle 1990s. And then coherent detection around the 2010 timeframe. And that's the two technologies that really enabled the internet to scale up in terms of capacity and so on. So, I'd like to say we're responsible for building the circulatory system for the internet. And that paid off tremendously during COVID because people could go and work from home — and there was sufficient bandwidth in most cases, and if not, there was relatively straightforward ways to add additional capacity.
What people also learned is that the architectures of old, where we typically had static architectures … and you would build a residential network and then in parallel you'd build an enterprise network, and then you'd build a third network for wireless. That wasn't the right way to do it anymore. We needed the whole infrastructure to be much more fluid dynamic. Adaptive is a way that we talked about it — building the networks to be adaptive, and that was a huge lesson, you know, learned about network architecture. But we felt fortunate that having developed these high-capacity technologies, we knew how to apply them to solve these problems.
Coyne: Before we started recording, you talked about the three C's capacity — no two C's: capacity and connecting, and then latency, which — as you said — has been late to the ballgame, ironically.
Alexander: So obviously, yes capacity is late to the ballgame in terms of a parameter that people really need to pay for.
Coyne: When it comes to connecting to the cloud, why is that so important for your customers, and where are we now and where do we need to be?
Alexander: Well, it has become much more important and a good way to think about it as we move towards a more immersive experience with the cloud. People talk about augmented reality/virtual reality, the need for [artificial intelligence and machine learning] to process things quickly — and having access to lots and lots of information like latency becomes more and more of an issue and it is largely ungoverned, and in many cases, even an unmeasured parameter on the network.
What I've started certainly to do is talk about mean-time-to-cloud, so from whatever device exists at the edge of the network — your smartphone or your computer, your smart television or smart refrigerator, your crypto miner, whatever it is — it has some expectation about its cloud experience. And when you kind of work through it and you put people in the loop, what it takes for them to experience a cloud correctly and what the machines expect when you're doing lots and lots of processing, what you find is numbers somewhere between sets of 5,10,15 milliseconds of governed latency is really what's required to make that work.
And then you back off and say, “Okay, what does that mean for where the cloud needs to be, or at least the first instance of cloud compute?’ and you're gonna find your edge cloud locations probably are going to be between 75 and maybe 150 kilometers away from you if you're really going to have an immersive experience.
So, if you, say, just to pick a number — 100 kilometers — and you go and say, in a major metropolitan area, how many of these things do I need to hit? You're gonna get a handful. And then for redundancy issues — what people like to call ‘blast radius’ — if something really fails from a major event, total power failure, earthquake, flood, fire, that sort of thing, the whole site goes down. How many people are affected? You factor in all the redundancy. You're going to be having dozens of these metropolitan areas to be able to, you know, provide that kind of connectivity and understanding how you do that interconnection, how you connect without enough capacity at low latency over to all the customers. That's part of what makes it fun is solving those architectural problems.
Coyne: So where do service providers play in this mean-time-to-cloud?
Alexander: What's really interesting because they are the ones that you know, bring traffic to the cloud and deliver it back from the cloud, is that most people see the cloud through a service provider. So, the advantage that service providers have is proximity, right? They're the closest to the customer. In many cases, they they're the ones that, you know, serve the traffic in either direction. And there's some kind of interesting use cases that sit out there around what is the right role for the multi-tenant, small data center out at the edge and potentially where service providers can be one of the use cases, which is admittedly very hard to solve. It’s kind of three-party problem, right? It says, you're in a smart city. That smart city has a service provider number one, an ISP, some cloud services that serve as a smart city. You have a vehicle coming down the road that has yet another second cloud service of some kind. And you have, let's say a pedestrian with its third. You're all kind of converging pedestrians coming to an intersection — and something has to change. Something has to intervene. You have to do something.
Does that happen locally? Which means it better be kind of a multi-tenant capability that's right there, and very low latency, or are we going to assume that all this stuff goes back up to three different cloud service providers, and back in the cloud, this all gets resolved? Most of the time when you work through the scenarios, if you let it go all the way back up to the cloud and then it has to do something, you can't intervene quickly enough. So, there's a lot of interest in figuring out, okay, if I need this kind of multi-tenancy for certain functions, out at the edge of the network in an intelligent environment, well, who provides that? A lot of people would say, well, that's the natural role for a service provider. They, in many cases have facilities already there. They're generally pretty well-connected. And so, that may be a very interesting role is they have that multi-tenant edge data center kind of capability that can bring all these immediate interactions together.
Coyne: How can they learn to take advantage of that? Are they set up now, like, from a skill set and a cultural perspective?
Alexander: It varies all over the place. They've got regulatory issues. They have business model issues. They have all sorts of things they have to sort through. I think right now they view it as an interesting opportunity. They're trying to understand, how big can it really be? What’s it really take to solve the problem? What kind of agreements do they need in places or a regulatory issue associated with it? So, there's a lot of, you know, complexity that goes into it. But from the kind of technology and architecture point of view, there's clearly a role. There question is, who wants to step up and who wants to solve it?
Coyne: You also talked about where business services are going — where are they going?
Alexander: Well, it's been an interesting bunch of conversations, and I don't yet have the right terms for it, but I started calling it the ‘yours, mine and theirs’ model.
The ‘mine model’ says, ‘Look, I'm a fairly sophisticated enterprise customer. I need a lot of connectivity, high capacity. I'm just going to get dark fiber connectivity from somebody. I'm going to give it to you, my folks, my tea shop, and my tea shop’s gonna run and it's gonna be mine.’ Right? And we started to see that certainly in the Fortune 10s, maybe Fortune 50s, who have the sophistication and no need for that kind of a capability.
The ‘yours model’ is one that's been out there for a long time — You go to a service provider and say, ‘You're going to build this for me. I’ve got 100 locations. You're going to get the SD-WAN. You’re going to get me the IP-VPN. You’re going to get me MPLS. Whatever it is, I want you to connect me up.’ Most everybody's pretty comfortable with that one.
The ‘theirs model’ is, again, one that we see kind of emerging. It’s a bit different. It says, ‘You get me to one of the, some of the major cloud vendors and once I'm at that location, they're gonna take the rest and do the rest.’ And what's been interesting – certainly from my perspective — is talking to the various service providers and the enterprises that have come through here. Not everybody is comfortable with all three models.
Sometimes they don't like the dark fiber models, maybe because they don't have enough fiber to begin with, or maybe the fiber isn't very well utilized. I've run into people where they've got 144 count fiber cables, and every one of them with gigabit Ethernet. There's probably good business reasons why they did it that way, but in today's world, it's not a very efficient use of the bandwidth on the fiber. I've got others that were very much about, ‘well, give me a connection to the cloud. But if all of the value-add is from the cloud, what's my role?’
It's been a very rich set of conversations around how this business and enterprise services space is going to evolve.
Coyne: Well, thank you so much for inviting me here today, Steve. It was a pleasure to meet you.
Alexander: And I can tell from all the tours that you took that you got the right sense of what we're doing.
Coyne: One last question – at the CTO level, what's next? What's on the horizon?
Alexander: Oh, lots of things because, you know, the demand for capacity and connections is not slowing down. If anything, the very rapid adoption of things like ChatGPT — which is the fastest technology adoption in history — that's going to continue this march toward 30, 40, 50 percent of your growth. The infrastructure that we provide is this kind of circulatory system of the internet, of the cloud, in some sense, right? That's enabling all this digital transformation. You can look towards it becoming effectively the lifeblood of multiple industries, all based on the kind of networks and connections and capacity that we can offer.
I can't say the sky's the limit because there's going to be limits to all this. But I look forward to the day when we can really say we've closed the digital divide; we can put people online anywhere they need to be; you can work for anybody from anywhere; you can learn anything from anywhere — that's priceless.
Coyne: All right. Sounds good to me, Steve, thank you so much. Thank you.