We had a conversation with Prayson Pate who is the CTO of the Ensemble division at ADVA Optical Networking.
ADVA Optical Networking is a telecommunications vendor that provides network equipment for data, storage, voice and video services. ADVA Optical Networking has a global workforce of over 1,800 employees and its Fiber Service Platform has been deployed in more than 250 carriers and 10,000 enterprises. The Ensemble product family is a truly open virtualization suite, providing powerful carrier-class orchestration, virtual network functions (VNFs) hosting, management and automation tools. It reduces cost and complexity through intuitive, intelligent and automated tools.
We managed to book some time with Prayson who is the CTO of this division, to talk about his background, SDN/NFV, emerging technologies and his thoughts on how Network Engineers can best equip themselves in the fast-evolving world of networking.
Tell me a little about your background before joining ADVA Optical Networking?
Sure. I was fortunate to get an opportunity to do some innovative work very early on, at a company called Siecor, and was fortunate to start working with some technologies that have lasted very well, and those are fibre optics, Ethernet and software. Admittedly, they’ve all changed a lot since a million years ago when I started, but the fundamentals are still there. After that, I made a pass through BNR which was part of Nortel. I got some big company experience, which I found I didn’t really care for too much. I then went on to a smaller company called Larscom, where I got some great experience selling to Telcos. I left Larscom in 2000 to co-found Overture Networks and that was a very formative experience for me. However, it was also tough because 2000 was not a great time to start a business in the telco space.
I bet, it must have been pretty crazy?
Yes, it was very difficult. We had the advantage that we were able to hire and keep some very good people because there was little competition. We were able to buy some very expensive test equipment on the used market, for pennies on the dollar. But getting those first sales was very difficult. Eventually, we did, starting with some smaller operators, like TW Telecom and WilTel and moving up to larger ones like Verizon. We did have a very good run there.
We were selling carrier Ethernet equipment to telcos and we started that right at the beginning before Carrier Ethernet had been defined. When there was this move from TDM services to Ethernet services, and the need to bridge the gap and provide Ethernet services over TDM, that’s the equipment we were providing. We helped the operators make that transition.
Did the dot.com boom and bust affect your business?
Yes, it did, because the landscape changed radically. There were a lot of companies that started up in the competitive operator space, the CLEC space that had initially gotten some funding and been through financing. A lot of those companies went belly up very quickly. The major operators went through a cycle where they were not spending anything, that put a lot of pressure on the telecom suppliers. Early on in the history of Overture, we had an offer on the table to acquire the company from of the larger optical suppliers, and then they retracted the offer and turned right around and sold themselves to another supplier. This was based on the doom and gloom that they could see coming from the telecom bust. It was very difficult to get started, but we had some good investors that had some patience and we some time to develop our product, find some opportunities, get those initial sales and grow the company.
For you, what was it in particular about ADVA, is it something you had a keen interest in? Or was it just an offer come at the right time?
It came at the right time, which was when ADVA acquired us. In 2012 we had made a change at Overture, moving away from our traditional business selling appliances to carriers, and into the virtualised software space. This was about the time that NFV started. We identified a good market and a good opportunity. The vision that we had is still the same vision we have today – so I think it was the right vision. Unfortunately, the market developed much slower than we anticipated, and our investors were losing patience, so they decided to get out. ADVA has specialised in identifying good companies that were available for low prices, and they’ve incorporated that technology and the people and have had a lot of success – both with their homegrown technologies as well as acquired technologies. It turned out to be very good from the standpoint of the employees, and almost everybody has stayed with ADVA. It was not a great exit for the investors, but they did get a reasonable amount.
Was there anything in particular about ADVA that you liked when they were looking to acquire you?
I knew a bit about ADVA because we had partnered with them over the years, but also competed with them in some spaces. What we did know going in was that they were a company that was highly respected by their customers, and highly respected by their employees. If you look at the various job sites, ADVA had very good ratings. ADVA has always invested a lot in developing forward-looking technologies, so we knew all of that going in. We have learnt that the view going in was the correct view.
Can you tell me a little more about your role and what your position is at ADVA?
Sure. At the time of the acquisition Overture had started down this path of virtualisation, and that product line we called Ensemble. When ADVA acquired Overture, they took the Carrier Ethernet equipment and merged it into their other product lines. They took the virtualisation software and put it into a new division called Ensemble. They did that so we could focus on building and marketing it as an independent software, decoupled from the hardware. Now I am the CTO of that Ensemble division. I spend a lot of time looking at future trends, speaking at conferences and writing blogs about our thought leadership. I work with customers at an executive level and help them understand what these new technology trends are, and how it can help with their business and achieving their strategic goals.
What would you say the Ensembles goals are for the short to medium term?
Good question. What we have done is operationalised NFV for this first wave of deployments, which are based on some current technologies like, monolithic virtual network functions, that are running in virtual machines and that are hosted on Intel-architecture servers. That’s a good start but that’s not quite the end goal that the operators want to see. The first thing we want to do is break up these monolithic virtual network functions into smaller pieces. Let me give you an example. If you look at something that you would call a router, it includes not only routing software, it might include some security functions, some firewall functions, perhaps some voice functions, some VPN functions, they’re all kind of lumped together. What the operators want to do is break that monolithic function up into smaller pieces so that they can assemble those pieces, as needed for an optimised and purpose-built service. That is sometimes referred to as a microservices approach.
The next thing they want to do is move from virtual machines, which can be somewhat heavyweight and incur cost in terms of computing memory. They want to move away from virtual machines into something called containers that are lighter weight. Finally, they want to move away from only having Intel architecture, to being able to support other architectures such as ARM or even serverless architectures, where they’re completely decoupled from the underlying infrastructure. It’s taking that first set of deployment based on today’s technologies and mapping them to the near-term changes.
How big is the team at Ensemble?
We have about 100 engineers.
Are they all based in the US?
We have engineers distributed between North Carolina in the US and Bangalore in India. We had opened an office in India with Overture before the acquisition, and ADVA has continued to maintain and grow that office.
Do you go there often?
I go over there around once a year, and I just got back a couple of weeks ago.
With the competition for you guys, and is there anyone in particular that you see as a big threat to you?
Yes, we see competition from three angles. First, there are the incumbent suppliers of networking equipment. When the operators invented NFV, they wanted to open up the market and create more competition. The guys who were the incumbents said they were going along with it but they’ve not been particularly happy about doing so. They are protecting their incumbency in their positions as much as they can. The operators are now being insistent that the suppliers open up their solutions and work in a multi-vendor environment, but they still are very powerful. The second is, there are some other players who like us want to address the true vision of NFV, in particular providing open NFV infrastructure. So far, we have been able to stay ahead of them, and we have more features, more performance and more wins, but they’re not sitting still so we can’t either. We have to continually add features and capabilities and drive our technology.
Finally, one of the ways I like to talk about the NFV is that it is intended to bring the power of the cloud to the telcos. That is related to things like multi-vendor systems and virtualisation and being able to work with new methods of development such as Agile and DevOps. That means the traditional cloud players and over-the-top players are looking expand beyond the data centre and push further out to the network. As we’re trying to take these cloud technologies from the edge of the network pushing in, they’re looking to expand their capabilities from the cloud and the data centre, out into the network. There’s a competitive threat from a number of areas, but that’s okay because as they say if there is no competition there is no market. Clearly, there is a market here, and we are happy to take on the competition.
With SDN/NFV when was it that you really first came about it?
We started hearing about SDN and OpenFlow in 2011, and that’s when it really started to be talked about as a next-generation technology. As we looked at SDN we thought it was interesting, but it didn’t seem to be what telco operators needed. It did seem much more suitable for the data centres, but in terms of providing traditional telco services and next-gen services, it wasn’t really aligned. We didn’t really spend too much time looking at SDN per se.
NFV was a different matter. Before the ETSI white paper came out in 2012, we started looking at virtualising the access network. This started with a slightly different model of virtualisation that the MEF were talking about with virtual NIDs. This was not virtualising in terms of virtual machines, but allowing the network interface device to have shared access between two operators. We were looking at that, and we were also looking bringing a more software-centric approach to how services were deploying the network. In fact, I gave a couple of talks about virtualising the access network at that time. When the NFV white paper came out in late 2012, we took that as a sign that the operators were serious about it, and we made a big bet on providing virtualisation software. We started by taking and virtualising our NID. At the time we were making hardware NIDS, and we took and converted that into software that would run on a standard server and started talking about the whole approach to virtualisation which we called Ensemble. We launched that in the spring of 2013. That was five years ago that we started down this path. It’s been a long slog, but we’re now seeing a lot of success and take up.
Especially in the last year or so, the last 18 months?
Absolutely. What’s really interesting is that while over the last few years you’ve seen some talk about NFV, a lot was lip service. Now you’re seeing the operators saying, hey when we said back in 2012 that we wanted to: embrace open systems, standard servers, multivendor and services built from components, new ways of working and new commercial models like pay as you go, we were serious about it. Now they’re starting to put their foot down, and all the RFPs that are coming up now are insisting on openness. They’re insisting on these new multivendor models, they’re insisting on new commercial models. That is great as far as we’re concerned because that’s what we’ve been saying, and that’s how we’ve built our product line.
Regarding the RFPs, who have you noticed, whether it’s an industry or a particular company, who have been the biggest adopters?
In terms of RFPs, every one we see coming out, we could’ve written it ourselves, because they’re asking for the things we are supplying. In terms of the biggest adopters, the ones who have been very public about it are people like Verizon, AT&T, CenturyLink and Colt. However, we’re seeing that even the ones who aren’t talking about it realise they have to move down this path and are figuring out how they can move forward. What is interesting is, the companies I just mentioned are the tier 1s. We’re now seeing a lot of the tier 2s and tier 3s that are realising that this new approach can bring them some real benefits. We’re able to now respond very positively when they say, we want to take and move away from the appliance model to virtualisation, we want to offer SD-WAN, we want to offer a virtualised approach, we want to get rid of all these boxes, we want to use standard servers. We’re able to say, okay, great we’ve got it, how many do you want? We’ve moved forward very quickly with them.
What about the enterprise space, is there any particular industry that you have noticed that has really picked up?
We’re seeing the enterprise impact from two standpoints. First, there’s the indirect. If you look at why the operators went NFV, it was to move away from appliances because if you have a model where services are tied to appliances, then every time you add or change a service, you have to add or change an appliance. If you think about a large retail chain that has 500 or 1000 or whatever stores around the world, and they want to add a new service, the operator has to send a person to every one of those stores to install a new box and that is not tenable. A sophisticated customer are now insisting that when they are putting out an RFP for telco services, many of them are starting to say, we insist that this service is provided on universal CPE in a virtualised manner. Because we want service agility, and we know if you come in here with a closed system, then we’re not going to get that – that is the indirect effect.
If you listen to people like Shawn Hakl at Verizon, they’re publicly talking about one of the reasons that they had to make a hard move into the virtualised approach, is because their customers were insisting on it.
Now we’re also starting to see a direct approach. Where we mostly sell to operators that we are working with a few retail, or enterprise end users who are looking to have their own virtualised solutions – for all the same reasons that the operators are doing it. This is because they have particular needs or very sophisticated applications that they want to host on this virtualised infrastructure. Between the direct and indirect there is a large pull from the end user space to move away from appliances and into more cloud-centric architectures.
With cloud-centric architectures, apart from sales are there any other big players that you see quite often? Not necessarily in competition with but that you collaborate with?
One of the interesting things about moving to a multivendor and open approach is that you wind up cooperating with people that you compete with in other spaces. I mentioned that one of our competitive threats was the incumbent equipment suppliers. As it turns out, we also cooperate with them. We host their virtualised network functions, their virtualised appliances on our infrastructure in some accounts, so it’s this new world of co-opetition.
Another thing that is interesting is that we are working with a lot of the traditional telco operators, but we’re seeing that some of our partners on the value-added reseller or integrators are looking to provide integrated solutions to the tier 2, tier 3 and enterprise customers. We’re also seeing some of the Managed Services providers who traditionally worked with the appliance-based approaches are now moving to a virtualised approach. Some of them are starting to provide integrated cloud/connectivity services, so they can provide an end-to-end connection with cloud services for hybrid cloud and multi-cloud applications. When you virtualise these functions you open up a lot of new opportunities for new service offerings.
With the way it is going with it is going are there any particular challenges you’re facing or that you will be facing the future?
There are a lot of challenges, and that is one of the reasons that it took so long to go from the vision of NFV to its growing implementation that we are seeing now. The first thing was cost and performance. Getting sufficient performance to meet the SLAs meant having a large and expensive server, of course of the cost model. Either you used a big server with plenty of performance, or you used a small server and you couldn’t get the needed performance. That was one of the first areas that we tackled by applying technologies like DPDK and software optimisation and working with our VNF partners to do the same. That was resolved a couple of years ago.
The next issue was commercial. Some of the early VNF suppliers were charging more for a software version of their product than they were for an appliance. They were doing that because the customers were insisting that they supply a software version, so they said, okay here it is, but we’re going to charge you more for it because we don’t particularly want to see it to you. That has also changed. Now you see much more realistic pricing models. Here’s a good example. Somebody might have had a virtualised appliance that was designed for a data centre and said okay, we can run this on our small server out to customer edge. But you can’t charge the same thing for a piece of software that is serving only one customer instead of a thousand. The market has become much more realistic, and the software suppliers are now pricing their software based on the value of the service that is being generated. Also, we’re going to models like pay as you go or on demand so that the operators are not paying for the software until they turn up the service. What they’re paying for is based on the speed or the number of users or something like that. That was the second challenge. The last challenge that we really focused on this last year is the operational aspects.
Just to go back on that, with the costing issue, is that because say if they were to sell a typical piece of equipment like an appliance they know that they would then receive funds from selling them and upgrading in a few years time. They would also get money from the maintenance of the equipment, whereas a piece of software is a one-off fee?
With our software, we sell it either as a one-time perpetual licence or as a recurring licence, depending on the model that the operator wants. In either case, if you look at the net present value of what we sell it for as a perpetual vs. recurring, that overall amount is smaller than if you’re selling an appliance. Because you’re only selling part of the solution. Part of the reason that the traditional players didn’t like this change was that they had a model selling a closed solution at price X, that could only be upgraded by them, and they could charge maintenance for it. They were asked to go to a model of selling a piece of software that’s maybe half X or quarter X. The amount of revenue that they’re deriving from it is smaller, and that is the big issue that they didn’t like. The problem is, they had competition from other suppliers as well as from open source solutions, so they’re forced to embrace of new models.
Going back to what you were saying, what were the last set of challenges for you?
The last set of challenges have been operational. How do you take something, that once you’ve convinced yourself that, yes you can meet the performance in the SLAs, and yes you can define a cost model to meet your business case? Once you have convinced yourself of that, how do you take this and roll this out on a scale, because with the telco operators this is all about scale and automation.
Over the last year, we spent a lot of time working with our customers, figuring out how to improve things like zero touch, security, authentication, fulfilment, licensing, management and performance monitoring. All these things are not directly related to NFV but are necessary to roll out a service in a telco environment. We’ve been talking this year, and just issued a press release, about a new solution release. The theme of this release is putting the M in MANO, where MANO is management and orchestration. There’s been a lot of focus on the orchestration side of MANO, and less on management. With this solution release, we’ve been putting in a lot of features and functionality related to the management. Specifically: how do you install it, how do you turn it up, how do you manage it, how do you upgrade it, how do you troubleshoot it and in short, how do you operationalise this exciting new technology so you can realise the benefits.
Have you got any success stories from that, where you’ve overcome the challenges from your perspective?
Definitely. When we talk to an operator, and they have started to look at NFV, and they have started to play around with it, they see on the one hand, yes there are some benefits that they would like to gain. On the other hand, they don’t know how to get there, and there some challenges they don’t know how to answer. When they come to us now, we say yes. Yes, we understand those challenges, yes we’re rolling those technologies out with other customers, and yes we can help you do this. This is real, we can set this up now and get you going. In fact, one of the services we offer our smaller customers is support in installing, integrating and even running it initially just to help them get going.
With the networking world as a whole, where do you see it going, how do you see it evolving further?
Now that we are expanding the cloud, connectivity is more important than ever. We’re ruled by our smartphones, and we expect high-speed internet connectivity everywhere – so that is really the next step. How can we assure that, wherever you go, whether in your office, your home or travelling, that you can have high-speed connectivity? Technologies like 5G are designed to ensure that high bandwidth connectivity is available for today’s applications, but more importantly, for the things that we’re talking about for the near term and medium term. That’s things like virtual reality, augmented reality, autonomous vehicles and IoT. All of those are predicated on having ubiquitous, high speed and secure connectivity, with low latency and at a reasonable cost. That is going to be the next challenge for the operators, building out that network to drive that connectivity, doing it at a low cost and figuring out how to make money doing that.
Are there any technologies that you’re particularly interested in?
There are a couple of things that really caught my eye. The first is not a technology per se. It’s this radical change in how easy to develop and deploy new applications and services. With the power of modern languages, software components, development frameworks, debuggers, open platforms in the cloud or on smartphones, it’s possible for a single person to develop and distribute an app or service to millions of users. It is possible for people who are not in the cloud business, like telcos, to move into that space and start applying these technologies to creating and deploying services. The whole barrier to software development and application development is so much lower than it used to be, that it’s unleashed all these technologies, innovations and opportunities that we weren’t even thinking about a few years ago. That is not a technology, it’s an intersection of a bunch of other technologies coming together to lower the bar. In terms of actual specific technologies, what I find really interesting, and a little bit scary, are things like, artificial intelligence, machine learning and quantum computing. The opportunity there is so large that it could really help improve our lives and unlock some interesting new applications. However, they’re a bit scary because we don’t really understand what they can do, and what all the negatives that are associated with them are.
Being a CTO, how do you keep on top of all the emerging technology and which one will be a game changer?
It’s very difficult because as you said, there is so much going on. I have certain websites that I go to, and I get certain alerts, and I’m a fairly good skimmer. I am able to scan over a lot of things and pick out what looks to be interesting. The other thing I am fortunate about is I have some very sharp colleagues and we all look at slightly different feeds and streams and when we get something interesting we all share it with each other. Even if I don’t catch something initially, it’s likely that one of my colleagues will.
With the websites, is there any in particular you recommend?
My two favourites right now are Light Reading and SDxCentral. They both have a lot of coverage of the NFV space, telco operators and have some good reporters and analysts that I have known for a while, so those are both good.
What has been your biggest achievements so far?
Well, I’m going to cheat and say two of them. First is the cofounding of Overture networks. As we talked about earlier, we started that company in a very difficult time and in a location in North Carolina where it was difficult to raise money, and with an equipment-based company, we had to raise a lot of money. We were in a business selling to operators who traditionally brought from very large companies, not small ones, but we were able to win business, and by the end, we were selling to just about all the top operators in the US and worldwide. I’m very proud of that, I’m proud of the team that we built, and the products we released, and the success we had.
The second one is related, and that is reinventing Overture. In 2011/2012 we decided to completely reinvent the company and make the transition from hardware to software, going to a brand-new space and defining brand new products. We were leveraging our existing knowledge in terms of the way telco services are built and the demands of creating software-based forwarding paths, but it was a net-new set of products and architecture. We defined that, and came up with a vision, marched down that path, and never looked back. It was a very good vision, and the things we were saying back then, we’re still saying a lot of them now, so I’m very proud of that initiative and that change.
I’ve heard some funny stories about particular vendors where the CTOs and CEOs are very against the idea of SDN, thinking it’s going to fizzle out, it is not going to be around in a couple of years to come. It must be a great feeling for you to say, this is what is going to happen and how it’s actually happening.
Yeah, we got out in front of it, and the only drawback is that it took a lot longer than we thought. We were overly optimistic on how quickly we could make that change but, it is happening now and we’re pleased with it. It is interesting that when we were reinventing Overture, I was also reinventing myself, moving from that inward-facing head of engineering to an outward-facing CTO. That is a pretty big change for a typical engineer to make, because most of us, like me, are introverts and are not that keen on writing and speaking. However, this role is something I have always wanted to do, and it has been very rewarding.
What does success look like for you?
I became an engineer because I like to build things. For me, professional success is planning and building products that customers want to buy and deploy because they support important services and they’re valuable to the customers. If I look back to BNR, the reason I left there was that the products I was working on got cancelled, so that was just not interesting to me. When I went over to Larscom, we were building products that were being deployed by operators. That was also really one of the rewarding things for us at Overture. Now, at ADVA we’re building these new and innovative products that operators want and are providing important and valuable service to them. That is what is important to me, and that is what success looks like. I take great pride in seeing these products being deployed by operators.
It must be great when you go to these events and people are talking about your products.
Yeah, it’s great when we see our ideas or our images or graphs show up in some of our customers’ presentations. Even if they don’t cite where it came from, we know where it came from, and a lot of the people do too.
From your position at the moment, looking out into the industry, do you feel like there is a technical or even software skills gap?
Absolutely. What you mentioned before about the network engineers, and how people expect network engineers to have some programming experience. Mostly we’re hiring software engineers, so they have the programming experience, but what many of them don’t have is the experience with the cloud, virtualisation and those types of technologies. In some cases, we are able to find the people with those, and in other cases, we had to look for people who have a good background, the right set of interpersonal skills and a willingness to learn. That goes back to one of the most important things for us when we’re interviewing is personal references. It is very difficult to understand exactly what somebody’s capabilities and desires are from an interview but, if you’re able to get a reference from somebody you know, then you have a much better chance of picking the right person.
When you say you look to see if a candidate has a good background, what do you class as a good background?
Most of the positions that we are looking for now are in one or two areas. Either they’re for the NFV infrastructure software, which involves real-time programming, Linux and virtual machines and containers and OpenStack and those kinds of things. Those are the ideal skills, but as I said, many times we don’t find those, so we have to look for somebody who has some good embedded programming, or some good basic Linux application development. That is one side.
The other side is our orchestration and management software that is more of a traditional application development. In that case, you are looking for people who have skills with working in large applications and user interfaces and access to standard components like, databases and development frameworks and that sort of thing. That is in many cases a little easier to find because there are more people who have those capabilities.
Are these all local to you in, these guys who will all be on your team in North Carolina?
North Carolina and Bangalore. We do development in both those areas. We have a sizable team in Bangalore.
With the skills gap, how do you think that the industry can address this?
Well, I think that over time these things tend to take care of themselves. When you have a skills gap, then there is more competition for the people, and the natural answer to that is that you raise salaries, offer benefits, training and those types of things. Then people will either move into that because they find it to be interesting when they’re studying as undergraduates, or people who have been in the industry for a while will change over and train themselves up, get those new skills so that they can move into that area. It is good to support that within companies. I was talking to an innovative telco that was looking to make much the same change, and when they were trying to figure out how they were to going build these new virtualised architectures the question was: do we take our telco network engineers and train them up in virtualisation? Or do we take these virtualisation guys, these cloud guys, bring them in and train them in telco. What they ended up doing was spending more time having their senior telco guys learn about virtualisation. While they may not have had the cloud experience and skills, they did understand: what types of services had to be offered, what the framework was, what the customers were expecting and in general, what was the answer that they were looking for, what was the problem they were trying to solve? Rather than focusing on the means, they were focusing on the end. That was very innovative and showing some faith in their current senior technical people. Not every company does that.
Another company that has been interesting is AT&T. They have made great public statements about their move into virtualisation, they said we’re going to hire some new people, and we hope that many of existing people can make the transition. If they can, we will help them, and if not we’ll find people who can. So it is a combination of supporting existing employees and bringing in new people.
How do you think we can get children excited about STEM subjects and moving into the technology space at a young age?
I have mixed feelings about that. On the one hand, I do think there are a lot of rewarding and lucrative jobs that need that STEM education, and from that standpoint, I think it is important to expose kids to that so they can see what is out there. For those who have an interest and an inclination, we need to do everything we can to support them and provide them with opportunities to embrace that. The other side is that sometimes we fall in a trap of trying to get people involved in STEM when that is not really their cup of tea. For example, my daughter very good at science and math, being the dad in the technology field, I was gently pushing her a little bit to see if I could get her to have a bit more of an interest in those areas. However, she just had zero interest, even though she was very smart, very good in those particular topics and very confident, she had no interest in going down that path. We just have to be very careful that we support people and give opportunities and make sure that those doors are open, but not push people into something that they may not want to do.
If you could give advice to a network engineer, what would it be?
We talked about the skills that are needed for network engineers. However, I would add one more thing, and that is understanding not just the means but the ends. When you’re working with a customer, or in your own network, what is it that you’re trying to accomplish? If you’re on the selling side or the supplier side, and a customer is asking for certain features, why are they asking that? Why are they rolling out this product or service? What can we do to make their job easier, or their services more profitable? Thinking about the larger questions and thinking about the context, is one of the things that separates the people who advance and expand their influence from the ones who tend to stay at a lower level. That would be my advice, to think about the bigger picture