 for tech and telecom professionals. And JSA Radio, your voice for tech and telecom, on iHeart Radio. I'm Jamie Scado-Pittaya, and on behalf of my team here at JSA, welcome to our monthly virtual roundtable. We are bringing together top thought leaders talking about topics important to our industry in our monthly virtual roundtable series. Available right here on JSA TV YouTube channel, as well as on JSA Radio. The only tech and telecom podcast series currently available on iHeart Radio. These monthly roundtables lead us up to our on-site CEO roundtables at our industry networking event, The Telecom Exchange. Next one up is November 14th through the 15th, 2016, at the Montage Beverly Hills in Los Angeles. More info at thetelecomexchange.com. Today's topic, SDN and big data, with a focus on practical usage and real-world implementation, has garnered a lot of support and social media buzz. Welcome, our live audience right here today, and thank you also to those who are watching on demand. This roundtable is brought to you on our JSA video platform, which allows our panelists to log in virtually from anywhere around the world. And today, we're spanning the globe. We've got boats logging in from Israel all the way to Los Angeles, and streaming live video feeds. Care of our partners, Pinnacle. So thank you, Pinnacle. And let's get started. I am honored to introduce our guest moderator and my longtime friend, Rosemary Cochran. She's the principal analyst and co-founder of Vertical Systems Group. Her expertise covers carrier ethernet, IP VPN services, business fiber deployments, cloud and data center connectivity, legacy services, migration trends, and fittingly enough, SDN and SC solutions, which makes her, of course, the perfect guest moderator today to talk big data and SDN with our all-star panel. Well, Rosemary, thanks for being with us today, and please do us the honors of introducing our expert panelists. Thank you very much, Jamie, and welcome to everyone who's watching. We're very pleased to have you join us for a very informative discussion about SDN big data. Each of our speakers will share their insight and perspectives from different aspects of this topic, and I'd like to have them each introduce themselves and share a little bit about what they do in the SDN big data environment. Bill, could you start, please? Sure. My name is Bill O'Brien. I work for Centrelink. I'm the director of the adaptive platform strategy and development team, and my team's responsibility is building out SDN and NFE platforms, and also we started one of the first efforts within the company around deploying large systems as well. Okay. Andrew? Thanks, Rosemary. Hi, I'm Andrew Kuzminski. I'm the COO and chief strategy officer for Perseus. At Perseus, I run our operations, engineering and product groups, as well as solutions engineering, and last year, I'm sorry, a little bit more than a year ago, we rolled out an SDN platform on top of our low-latency infrastructure to service the financial services industry in a unique way. Thank you very much for having me on the panel today. Okay, Steve? Hi, my name is Zev, and I work for MRV Communication. I'm the leader of strategic marketing as part of my responsibilities. We actually doing research on SDN and NFE and implementation as part of our products, a packet optical solution, including software orchestration, and addressing different parts of the use cases, including data center, interconnect, bandwidth on demand, virtual CPEs, and eventually the whole different variants of implementation that's supposed to deliver, instead of big and down pipes, much more intelligent and dynamic network. Thank you very much. And just to start, we'll talk about SDN. Software-defined networking is, as we all know, transforming the telecom industry. We're moving toward more open, more scalable, more flexible, and more data-driven services and solutions. Today, we're talking about the big data piece of it all and what's happening in this transition, specifically on use cases, implementations, on new services, on new deployments, as well as the challenges. But I'd like to start with the architecture view and the platform level. Bill, you've talked about, described, integrated metadata as the oxygen for SDN-based service platform architectures and have, obviously, lots of experience in development. Could you talk a little more about what that means and how and why that's so essential to the transformation? Yeah, absolutely. I mean, when we started this journey back in 2012, we realized, you know, with these highly automated systems that come as part of kind of the SDN and NFE architectures, with all that automation and additional layers of abstraction with virtualization and so forth, that we're going to need to capture a lot of data. And before, typically, systems would be staged. They'd be really separate from the platform itself. Maybe there'd be a solution that had integrated analytics, but everything would have to kind of go back to a separate system. So early on, we kind of conceived that the platform itself should really solve for a lot of these problems, should inherently have some analytics capabilities and, more specifically, the ability for these feedback loops. So with the robust set of APIs that are typically inherent with any SDN and NFE platform and the automation that goes into play, we begin to ask ourselves, you know, how can we even automate typically decisions that are made by humans and then use this and build intelligent feedback loops back into the platform so that the platform can adapt to the operational, you know, kind of behavior of the platform itself and all the dynamics that happen with that, as well as how individual users or customers are using our network services here at CenturyLink, you know, from a range of IP and PLS to Ethernet-based services. So there was a need for kind of building out a big data platform leveraging, you know, a lot of the typical tech tools that we see, you know, between Hadoop, Spark, and using various classes of machine learning and bringing data scientists on board to accomplish that. Excellent. And that's really a good intro then, Andrew, to talk about, you know, with all that behind the scenes and this type of amazing platform SDN-based with big data. What types of new service offerings is this enabling in terms of, you know, analytics for your customers? Sure. Thank you. So there's definitely a paradigm shift that's occurring, especially as it relates to the financial industry, the financial services industry as we're servicing those customers. And in order to keep up with that shift, you know, we've adopted SDN as the platform to be able to service those customers in a much better way. So, you know, we've called it, you know, the NetJet model or this uberization of connectivity where, you know, things have changed and the power of the network, where it can go all the way into the application's hands, at least at first needs to go into the customer's hands. So that customer enablement, allowing them to log into a portal to order their own services is at least the first step towards offering those customers new services, whether that means gaining access to new network routes or new compute virtualization locations. You know, that customer enablement is that first criteria. So I guess the second one is, you know, as a part of that is real-time delivery of services. So, you know, the purchase SDN platform needs to enable our customers to be able to operate more quickly and efficiently. And I have actually a good real-time example. This happened just two days ago. And the emphasis can be different in very cases, whether from a financial services firm strategy changes or there's a network event. In this case, there was a network event between the Hong Kong Stock Exchange and the Japanese Stock Exchange, where there was a cable that was going down and a cable that was cut at the same time. And our client needed to be able to turn up new capacity between those two points. By logging into that, to the first SSDN fabric, they were able to enable their services to go live within 24 hours, which given that there were no cross-necks in place or anything of that nature, there was still some human interaction. But enabling those customers to turn that service up that quick is a very powerful tool. And then finally, I think, you need to think about, well, from the person's perspective, right-sizing and scalability of services. So, you know, this particular client or any other particular client that's using the first SSDN fabric might realize serialization issues or propagation issues as a result of congestion and they can alleviate those by adding capacity or potentially they realize that they can be more efficient and less capacity. So, giving our clients the opportunity to upgrade and downgrade capacity is a critical component of the first SSDN platform. And while we definitely have other products and services that are being added to this environment, that was the starting point for us. That's, you know, a very good example. Those types of use cases of what has to happen and the ability to really enable that functionality that quickly, which brings us to, you've got kind of the, you know, the fuel of all this to talk about some of the solutions that, you know, the big data is, you know, kind of helping with and how that works in terms of going forward. Yeah, so, actually, it's interesting because obviously the, you know, the linkage to big data is trivial. We are sort of a data-driven society that expects digital experience and we perceive the futuristic network that's supposed to respond on demand to our requirements and really simplify and literally use those unlimited pull-off resources. It's actually interesting if you ask a software guy, the network should be unlimited in terms of pull-off resources, which brings again into the concept that drives the software disruptive networking. Yes, it sounds a bit erroneous, but this is literally of disruption, a story that will shake the industry that tend to emulate sort of a cloud modeling that really speak about large scale of volume, velocity and variety, which basically this is the big data. So if we judge the numbers of data flood and Rosemary as an analyst, you have really good insight on this. It will continue to evolve from IoT and the mobile devices. Commercial mobile devices tend to go into the range of more than 10 billion in the future and if you look on sensors that bring also a lot of data, so the estimation of 30 billion of RFIDs or later any other aspects of artificial intelligence will bring the big data as a big topic. So big data creates sort of tectonic shifts that go hand in hand with server virtualization and new traffic flows in the network that differs substantially from traditional client-server model. Due to the elasticity nature of cloud environment, big data processing reflects server-to-server modeling with traffic load that change in location, in intensity like Andrew just mentioned over time, which practically demanding a flexible approach to managing network resources. So this means that the lever of new services over the network and adjusts performance and service parameters on per session, per application basis is simply time consuming and not efficient with all the processes and technologies. So this is supposed to change with SDN paradigm. The objective of SDN is really to enable as discussed transformation to open programmable and virtualized networks in application driven. So the new approach is really kind of to empower the software to tell the network what we want it to do and configure all of resources as opposed to how to do and configure manually each device in the network to set up end-to-end service to business customers. Additionally, central software intelligence should enable collection of analytical data from entire physical and virtual networks to keep real-time snapshot of capacity usage, SLA performances and highly granular data on application and user experience. So practically as of today, the SDN enhancements in telecom infrastructure and telecom apparatus networks have enabled some telecosters to accelerate and simplify provisioning of their own bandwidth and data center resources. But you see it from our experience, the next stage of evolution is that business customers now would like telecom apparatus to extend this virtualization of the network across their wide area network to deliver new and more attractive services at a lower price, obviously. And so this will force new business models of the option to reflect sort of a win-win situation of business customers, but still enabling the telecom industry to keep the pace and not move to chapter 11 in the challenge. Lots of challenges. That brings us to challenges. One of them is, and that was an excellent view of the different pieces of the network and how we are expecting more interaction, more of the data-driven kind of experience. And the telecom world, one of the big issues is the connectivity piece of connecting, particularly as from a service provider standpoint, you're not really extending your footprint everywhere on net, and that interconnectivity is really a challenge. Maybe, Bill, you may speak to that to some extent in terms of how that is changing or how the data analytics will help in that respect. Yeah, I mean, I think there's some opportunities there for analytics or the data that we can use to improve SLAs or drive new connection models. So, I mean, specifically when we're capturing this data, you know, the opportunity to improve, be able to predict, really, going to a prediction models of knowing the behaviors of certain interconnects and being able to mitigate that fairly rapidly is going to become more and more critical, right? So, I mean, really, the driving motivator here is improving the experience. And when people are buying telecom services, they're really looking for super high availability and performance, and those are guaranteed with various SLAs. So, by having, by taking all those metrics that you can start to build in algorithms that begin to let us know, you know, potentially partners that are at a higher risk, you know, when particularly you generally see trends around failures and around load saturation so that we begin to mitigate that ahead of time. Great, great. And that brings us to maybe, you know, talking about some of those connectivity issues. Andrew, with clients all over the world, are you seeing different, you know, kind of transition rates for your customers in different regions? Or is there any opportunities that you see on a geographic basis that are more diverse? From a client adoption perspective, you know, we're seeing a huge uptake in the client base in Asia, specifically. That entire region seems to be adopting the platform much more quickly than North America or Europe is. You know, there's, you know, I think with the adoption of a new platform or a new technology, we constantly get asked by our clients, you know, why SDN? What is SDN? I mean, at the end of the day, our clients, our financial services firms, they're not, I mean, some of them are technology firms, but a lot of them are traders. You know, they're trying to figure out how to trade efficiently on different marketplaces. So there is definitely some fear in the adoption of a new platform. And there are a lot of questions, but there certainly are a lot of benefits to the platform. And, you know, from a regional perspective, we're seeing less, I guess, less fear as it relates to adoption of the platform from our Asia-Pacific customers. I just wanted to touch on something Bill said before, which is, you know, predictability and, you know, building a predictive network as well. And I think that's one of the key elements that just needs to just be re-emphasized because that is one of the huge benefits for our clients as it relates to them being able to adopt this service and having software to be able to create those predictive analytics so that we can create a more efficient, more effective network that is as scalable as a network should be to react to what Zeve said before, which is the application, which is going to be ultimately driving those network decisions. So, yeah, sorry, I'm just digressing a little, but I just wanted to touch on those points to re-emphasize a little. I think that's an excellent point because, clearly, predictability and performance are key for all of this, too. The work would be successful. That is one of the major goals. So, when we talk about challenges, let's you talk a little bit about that, so there's the challenges going ahead, what's required, what's going to be the focus of what people have to deal with in this transformation? What would you say is kind of the top thing there? You know, for the telecom market, the biggest obstacle is to complete end-to-end service concepts and do it in de facto technology and industry standards. This includes network provisioning, data center resources, application management, user portals that should complete the change of the ordering model and align with cloud business concepts. Let's not forget, also, the telecom environment is typically more conservative and usually telecom builds everything twice for very high SLA as mentioned previously. And I even allow myself to use a specific quote that was said by a century-linked CTO, Amir Hussain, and he said, we need to transform from telephone company to an IT-based service company, which is really mindset shift from the old style of telecoms and think more software. Now, the other side of the story is that the biggest meat of SDN is that it's one size fits all, but this is a longer transformation with processes and there are no shortcuts. And in fact, virtualization in data center is mature. I think the use cases we discussed previously based on Bill and Andrew are unknown, but the existing telecom, weather and networking and internal IT are completely different in terms of technical requirements, adoption process. So challenges for mass market implementation still focused on requirements that tend to be the silver bullets of such radical transformation process. Just if we look at the modern OSS that's supposed to happen with the shift that's supposed to happen in standard APIs for automation and abstraction of the network, this is still evolving stage. Some companies, some telecoms, they're already in more advanced mode and we heard a lot about AT&T domain to the toe project, but many others are still doing wall-garden, multi-domain environment use cases and analysis. If we look at this stage, more IT transformation and cultural mindset will need to change because it's really knowledge, learning and corporate strategy that's supposed to happen. Bill, you have anything to add? You've made a really, really good point. I think one of the major challenges are going to be for most companies is going to be the cultural, right? Getting, there's organizational changes, the procedural changes that are going to have to take place. And as you pointed out, kind of end-to-end, that's impacting a lot of desperate systems. So here, we're kind of really making this mind-shift change where we're talking about kind of federation of internal platforms making that shift to DevOps and this highly integrated data model that now we're having systems and machines make decisions that was really the domain of humans. And that's a huge shift. And so there's a trust layer that needs to happen. And we'll see that come in with vendors where they're going to be having this integrated, kind of smart systems. But in these large heterogeneous environments, we can't really rely on a single vendor to provide kind of those smart solutions. So we need to look at it holistically. And so I think individual companies will need to build up, their capabilities and kind of this data science in order to allow for this transformation to happen. And that really requires that platforms are put in place, not only kind of these automated SDN, but really understanding what big data platforms look like and what they can do. It's not trivial. And you have to put in the time. I mean, we look at large web firms. They've invested almost a decade into this. And they're doing very, very interesting things now. So this is not something that you can necessarily buy completely off the shelf. You're going to have to make that investment in your organization as well as in, you know, the knowledge within your company so that they can begin to evolve in the development of the algorithms to make the smart systems and smart networks. Right. And I can, I would say that that is clearly one of the things that we see from tracking of this development and to move to SDN. Platforms and it's not just a technology issue. And clearly it is a challenge. It's going to be bought off the shelf and it does take time. But there is the other corporate kind of cultural issue, the commitment and people. People are a big piece of that in the skill sets. So there's a lot of aspects that are yet to be achieved. So we've got maybe one more, you know, question to, I'm going to throw to you, Andrew, about challenges, you know, we talked about some things that are happening on the services side and the, you know, companies that are trying to deliver services. From your customer standpoint, what are the challenges that you see for providing services to customers or from the customers from what they're telling you about challenges in moving ahead? I mean really the biggest challenge has just been, has been the risks of, you know, fear of the unknown. I mean, the customers are happy to get on to a service that is more secure, that is more scalable, that has, you know, better availability as a result of the predictive analytics that we're putting out, that we're utilizing. And they're, frankly, they're happy to migrate away from legacy extranet services and, you know, things like MPLS and get on to a smarter infrastructure. I think it's, you know, fear of going on to a new platform especially from a financial services perspective where they're unaware of kind of what they're getting into is always a little scary. But, you know, for Zeve's comment and Bill's comment, you know, there are no shortcuts to building a network like this. And as long as you don't take any shortcuts and you hire the right people that have the right skills and you go through the correct processes of building the correct inventories and knowing what you have and doing it the right way, then you can overcome the, you know, the objections that any clients will have. And we have a lot of clients that are already using this platform very happily and very successfully. So while there are challenges, they're not overcomeable. Well, that interacts up, Jamie. I'm going to turn that back to you, but thank you for a great insight from our panel. And thank you, Rosemary, for moderating our roundtable on SDN and big data. Thanks also, as Rosemary just said, to our esteemed panelists, Bill O'Brien of CenturyLink, Zeve Dre, MRV, Andrew Kisminsky of Perseus. Thanks for your thoughtful insights on the SDN challenges and opportunities that lay before us. Thank you audience for joining us. If you want to see this and other monthly virtual roundtables on demand, plus the calendar for upcoming roundtables, both virtually and at Telecom Exchange, go ahead and check out jamescato.com and vtelecomexchange.com. And if you'd like your C-level featured here next time, go ahead and email us pr at jamescato.com. Thank you everyone for tuning in to JSA TV, the newsroom for tech and telecom professionals, and JSA Radio, your voice for tech and telecom on iHeart Radio. Until next time, happy networking.