 Hi, I'm Peter Burris, and welcome to this cute conversation. I'm being joined today by Pensa, who's a leader in the network virtualization world. And specifically, from our Palo Alto studios, we've got Udla Settler, who's the CTO and founder, and Steve Dyche, who's a VP and general manager at Pensa. Now guys, we're going to talk today about this notion of network virtualization, tying it back to what's happening within business, why is it becoming a crucial technology to invest in and how it literally becomes an essential element of a digital business. But let's start with a little context. So if we think about what businesses are trying to do today as they go to a world that is more dependent upon digital technology and digital transformation, they're literally trying to apply their data in new and different ways, which requires that they connect data as an asset with people, with customers, with other business entities, much more successful. Now in the past, we kind of knew where the asset was, we knew where the application was, we knew where the person was, mobile made a little bit more difficult, but now we're talking about literally absolutely requiring the flexibility in how our network is set up if we're going to take advantage of the opportunities that are presented by digital business in real time, as well as attend to some of the threats, whether they be secure or otherwise. Take us through that, why don't we start with you and take us through where you think this has started and where it seems to be going. Yeah, absolutely, thank you Peter for having us on here. So as you mentioned, the world is changing, the way people interact with each other is changing, the way people acquire things is changing, the way people sell things are changing, the way people view things are changing, so essentially consumption models are just being turned on their head, right? So that really has a tremendous impact on the underlying infrastructure. Not just the network, essentially how compute works, how storage works, and fundamentally how network works. Networks are very, very critical, right? They're essentially the circulatory system of the beast, right? And you know, it's the arteries and the veins and you can't have a fixed system that essentially was put together 30, 40 years ago and that is expected to meet the demands of how things are being done today. So fundamentally things have to change, not only from a business point of view, from the carriers have to change because they cannot be just a dump pipe anymore, they have to get in on the action of who's producing things and who's consuming things. So there's a lot of business drivers and there's a lot of technical drivers as well. So Steve, let's pick up where we were just talking about and describe some of the challenges of the carriers in particular phase. So it used to be you'd negotiate for circuit, you'd pull the circuit, you'd implement the circuit and it would just be there for an extended period of time. But now we're talking about something that requires a lot more flexibility, a lot more agility, a lot more plasticity. We like to talk about plastic infrastructure. It's capable of both scaling and accommodating new workloads and workflows much more faster. How are the carriers trying to adjust to this and how is that catalyzing demand for new types of technologies? Right. So if you go back to what Ujwal was saying about the changes in the external environment what you were mentioning, whether you're a business or a consumer, consumption models are changing. The fickle nature of individuals is changing. Digital transformation of corporations is driving new ways of thinking from a business model perspective, which leads to IT needing to be an enabler. Not only supporting the business, but driving the business. And that puts a lot of pressure on traditional suppliers of services, that including the communication service providers or the telcos or the carriers. The carriers have a number of challenges and this is not new. This is not new by any stretch of the imagination. They're under enormous pressure based on that fickle nature, based on the speed of business, the speed of innovation or the need for speed of innovation, cost optimization, and et cetera. And they're seeing massive demand from an apps perspective, from mobility perspective, from data consumption perspective. Couple that with enhanced competition. And this has been going on for a while, including the over the top players, the Google, the Facebooks, the Twitters, the Snapchat, anybody developing content is a competitor now. And then add to that the traditional environment or the infrastructure that usual references is highly rigid. It's traditionally been a very rigid, siloed, proprietary type of infrastructure that the carriers utilize to implement the services that they deliver to a consumer or an enterprise. In today's world, where it's all about agility, flexibility, and speed and cost optimization, a rigid, siloed infrastructure is not going to get you where you need to go. And it also gets prohibitively expensive because you want that flexibility in your deals and your pricing and whatnot to move traffic where it needs to. Yeah, in many respects, it's obvious that the telecommunications companies are still reeling from the 1990s losses of data comms as a major, major center of IT organizations. When we all went to IP, that led to some very, very different approaches to doing things. Take us from that moment, if you will, of, well, suddenly communications went all to IP, voice over IP, et cetera. How have we progressed to the point where now we're really in a situation where we can completely virtualize the network? Usual, why don't you give us some background on that? Sure, you know, as you said, the service providers got a bit of a knocking back in the 90s and early 2000s at a big telco bust of early 2001 and a lot of people got caught up in that. Well, it has moved to IP, but it's still very much in, you know, the assets are being locked or the potential is being locked into, you know, physical assets that cannot be moved. Like, router, switchers, load balancers, firewalls, all sorts of stuff, right? That is critical in a proper functioning of an application. At the end of the day, infrastructure exists to serve an application, right? The demands by an application on its underlying infrastructure is constantly changing. Now, the application architectures are evolving. Microservices architectures are being finalized. A lot of, so they essentially have completely different demands on what infrastructure needs to provide them. So now that we have virtualization technologies that initially came to compute and then storage, and now, you know, I like to say that networking came a little late to the virtualization party, but it is there now. Now, there are technologies now that are available, that are being developed. It's still, you know, work in progress, but you can finally transition from a fixed physical asset base to a more fluid asset base, right? Software base. Software base, right? You called it plasticity, right? Actually, it's an excellent term, right? And we call it dynamicity, right? And things are moving around. So the line between applications and infrastructure now is blurring, right? What is a virtual load balancer or a virtual gateway? It's an application running on commodity hardware, right? So it is, you cannot think of it in the old terms of here's an app and here's infrastructure. You really have to intimately tie them together and that's where the world needs to go. Well, I want to pick up on what you said about, we've been able to virtualize storage, we've been able to virtualize service for a while, and network you said is late to the party, but let's explore why network is late to the party and why network function virtualization, whole entity thing is so essential. The network is what connects all these things. And so it's almost by definition, it's a presumption that I'm not taking the common set of storage from a single place and putting it together, and then virtualizing it, it is literally, I got network assets all over the place, how am I going to commonly virtualize this thing to improve the flexibility and rigidity, again, the plasticity of my business. So what types of challenges does a decision-maker think as they have to go through, as they think about what it really means to design, envision a more virtualized network? Because it's not like designing or virtualizing a set of boxes. It literally is virtualizing something that's quite different. Absolutely, you hit it on the nail on the head there. Virtualizing compute and storage while extremely challenging is not quite as challenging as having to virtualize the network, because the network is the one that actually computes, connects all of these things together. So if you make one change over here, it'll have a ripple effect somewhere else in the system. So the challenges that a designer would face now is I can't do it all in one go, right? I can't throw out what I have that I built up and that I quote-unquote perfected over the past 30, 40 years and then start from scratch again. I have to pick and choose my battles. Where do I go? Where do I start virtualizing? How do I connect that to my existing physical infrastructure? So how do I do this in a manner that allows me to do this in a transition that is not very bumpy, right? That allows me to start virtualizing parts of the network, the core will likely not get virtualized anytime soon, but it'll start at the edge, right? The other set of challenges now is things that have been challenges even in the physical world, but there's a lot of maturity there, let's call it that, right? Performance, security, right? Resiliency, those things have been beaten on for decades in the physical world. And, but you really literally have to go back to the drawing board as you're doing this transition from physical networks to NFV. Now, where are my packets going? How do I debug? Fundamental questions like, how do I trace my packet, right? Which packet carried my virtual load balancer? If my virtual load balancer is being co-hosted on a commodity hardware that is running something else, how will the chatty neighbor issue affect my load balancer? How will the chatty neighbor issue is? Is chatty neighbor issue is something that I don't even control, right? Well, this is basically concerns, as you said, that when you start putting networks together, this device or this thing is going to affect that thing. Absolutely. If this thing starts chatting, it's going to have a potential impact on how that circuit or that device behaves in a broad network context. Exactly. So, not only do you have to revisit challenges that you unresolved over the past 30, 40 years, a new set of challenges have emerged as well, right? You know, challenges that, you know, virtualization is not a simple thing. There's a lot of complexity involved, a lot of new technologies, a lot of new protocols, and all of new standards are being written, right? So, you not only, so you have to not only replicate and evolve your networking function and your networking services, you can't lose anything in that transition, but you also have to make it programmable, elastic, and all the goodness that virtualization is promising you. So, there's a whole set of challenges looking at us in the face. So, Steve, that means that an organization that wants to undertake this challenge has to think differently, but also has to do differently. So, historically, we've had a group that did network, and we had a group that would network administration, server administration, storage administration. And now we're talking increasingly about these groups coming together and having to common practices, sometimes different tools, but nonetheless, presume a degree of convergence that we didn't have to deal with before. The storage service guys are still having problems with that. What do the network guys have to do differently to truly take advantage of some of these things? Well, I think the challenge is not unlike what we're seeing philosophically of when companies try to move to the cloud. And the technology piece of moving to the cloud is probably the least critical. I would agree with that. Honestly. It's a three, I look at it like a triangle. There's the technology, there's the people, and there's the processes. And one of the fundamental elements of the cloud or anything we're working into to gain greater, as usual mentioned, the complexity to deal with is automation. You can't throw bodies at this anymore. The network is way too complex. The IT infrastructure is way too complex. So when you look off the technology and look at the processes, what's the one thing that you need to do before you start looking at automating a process? You've got to standardize it. And to your point, you can't standardize a process unless you have all the groups on board thinking the same way, approaching a problem the same way and trying to solve the problem the same way, and a governance or interaction that aligns to this new process. And very importantly, it presumes that that is an asset that is worth governing. Correct, correct, which it may not be, but you have a tremendous amount of legacy infrastructure and you have a lot of new infrastructure. Well, when we're buying hardware, it was pretty obvious, but now that we're doing things in software, people sometimes go, oh, I don't know, is that really an asset? It becomes an even more important asset to the business, and therefore the governance of that becomes even more important. It becomes, and then if you take it to the third prong of the triangle, the people. This is a different, this is the exactly the same mental exercise that companies are going to as you move to a hybrid cloud, on-prem, off-prem, and the look at, do I have the right skill sets to be able to do this? Do I retrain? Do I hire from outside, understanding that these individuals are very hard to come by? There are shortages and they're extremely expensive. Those same challenges, probably multiplied by decades of legacy inertia within a carrier, creates a set of challenges that they have to deal with, and they have to deal with it urgently because there are things on the horizon right now that Usual was alluding to, that make what we're talking about here around network virtualization or the virtualization of network services mandatory. This is not an option anymore. If you looked back a decade or two of the way the telcos or the carriers were evolving their business model and their technology, we've come to a point now where questions about where it's going aren't really necessary anymore. It's, we now need to move or we have an existential threat. I want to build on that because I think one of the reasons why it becomes so absolutely essential is unlike 20 years ago where you could get it right and then plan for how it was going to change over time, in today's world, getting it right on day one in a way that isn't flexible is guaranteed to be wrong on day five. And I think that's one of the reasons why this whole notion of virtualization becomes so important. Another thing I want to mention very quickly is we've done some research at Wikibon on what we call the iron triangle problem, which is that you've got tool vendors, hardware vendors, administrators and automation all having different ways of doing things and that's a hard thing to break, especially as you start thinking about going to the cloud. So talk to us just briefly about how network virtualization technologies facilitate that journey to the cloud. Yeah. Let me take a first stab at them while we're on it as well. So as we talked about, the infrastructure today from a telecom carrier is basically very rigid. It's siloed. There are siloed to deliver a specific function or service. Firewall, router, switch, intrusion protection systems, packet gateway, et cetera, et cetera. As the need, as the business model pressure or the business transformation pressure comes down in order to be more flexible, agile, innovate with commercial models, engage in new types of business. The current network infrastructure can't handle that. It can't flex. You have to basically, it's the typical procurement model to go out and get when you want to launch. If you want to launch a new service based on the old model, it could take you a year. That's an highly expensive, takes you a long time and it's inherently risky. Who knows what's going to happen a year from now as you try to launch something. So we want to diminish the likelihood that we have to make a change at the hardware level. Correct. That's basically what we're trying to do. You basically want to have the flexibility to actually fail. Well, not have the hardware fail, but you want to be able to make a change that could be wrong. Correct. So keep so that the hardware can be stable, but we're introducing a degree of flexibility above that so we can make the changes that the business needs in software. Correct. We're not having to renegotiate a whole bunch of stuff. Correct. So what you're doing with network virtualization is you're taking that siloed model, which is a tightly coupled software function, a router on top of proprietary hardware, and we're decoupling the software from the hardware, and we're transitioning that, we're either rewriting it, we're lifting it, we're manipulating it, maybe it's now cloud native, and we're putting it on a virtualized environment on top of industry standard hardware, which gives you agility, flexibility, and cost savings down the road, more likely from an operational perspective as well. But still invoking the stuff that we know works is already in place, so we don't have to replace the whole thing. Correct. Now you've got a software horizontal platform as opposed to vertical silos, where you can quickly instantiate services, and if they fail, or from a business perspective, I can remove them and utilize that platform for another service. And very simply, the operating model that is suggested by that software orientation is the operating model that is associated with cloud and cloud services, and by doing this kind of thing in the network, it provides a very, very simple onboard to the cloud. Absolutely, I mean, it's essentially the same thing, right? I mean, something has to run it, so hardware is necessary, so it's very, very critical to the equation. But the logic, let's say in the network, the switching logic, the routing logic, the security logic in the network, doesn't have to reside in hardware, that can be implemented in software. That needs to be implemented in software, no. Applications have different needs. What happens, what will happen if a new Facebook comes out tomorrow, right? It has, it'll have very different networking needs from what you have today. If the requirement is there to go replumb a physical set of infrastructure, it'll fail, right? So you need to be able to layer on your required infrastructure. Yes, the commonality of the hardware will, it'll provide you certain set of capabilities, but the logic, the business logic of the infrastructure, it will be implemented in software, and that will be layered on the hardware. So we've talked about a lot of different things here. We've talked about digital business and how it's driving the need for greater rigidity, what I call plasticity in the infrastructure, how a software approach, software-driven infrastructure approach is crucial and now is available in the network. But every company has to make some decisions about how its product or its set of capabilities are going to map to the priorities. It's a highly dynamic, transformative environment. As thought leaders in this industry, what choices did you make when you were designing your product and what are customers getting from those choices as they think about who to work with and how this is going to impact their business? Yeah, I mean, one of the first things that we looked at was that we also needed to be cloud native, our architecture, what we do needed to be microservices-based. How we deliver that is a different issue, but the fundamental architecture had to be cloud native so that we are in the business of helping NFE technologies to get to market so that apps can get into service quicker. So we ourselves need to be aligned to that. The other things that we looked at is, okay, which part of the market do we want to address? Where do we want to start? So we want to start, look at segments in the network that are more mature for virtualization than others. Right, as I said earlier, the core will likely not get virtualized anytime soon. But at the edge, there are a number of use cases at the edge that we want to focus on. Just one or two. Use cases could be as divine as a use case for NFE and there is virtual CPE for mobility packet core, which is actually a fundamental requirement for 5G architecture being rolled out in the next five years. So those, and some of the engagements that we have right now are actually focused heavily on the mobility aspects of it. Virtual CPE is another architecture. So those are all use cases. Yeah, those are important. Those are important. But I mean, following up on usual, I think the choices made by Pensa in particular, both from an architectural perspective and where the use cases are, it was also important to identify the areas where we could make an impact. We could create differentiation and solve some pretty tough problems that everybody recognizes as a problem in a very chaotic, evolving market where standardization is still in its infancy. So if you look at what Ujjwa was saying, at the end of the day, what we're trying to do is, it's quite challenging with everything that we've talked about to bring a network service, a virtual network service to market. Design it, test it, validate it, build it, deploy it. What we've focused our attention on is a lot of that good stuff right in the middle of helping people along with intelligent automation and modeling of helping software vendors, service providers, SIs, network equipment providers, whoever is developing, porting, migrating, a network function into the virtual world for deployment in an NFV environment. We've helped them along that whole value chain of building on to make it simpler, accelerate it, and ensure that it's correct. And over time make it cheaper. Absolutely. All right, guys, this has been a great conversation. Thank you for your time. We'll talk about the evolving world of network virtualization with Pensa. Specifically, we've been here with Usual Settler, who's the CTO and founder, Steve Deitch, who's the Vice President General Manager. Once again, this has been a CUBE conversation from our beautiful studios in Palo Alto, California, talking to Pensa about network virtualization. Thanks for joining us. Thank you, Peter. Thanks.