The following transcript has been edited for length and readability. Listen to the entire discussion here on The Broadband Bunch.
In this episode of the Broadband Bunch, we speak with Robert Varga one of the early and major contributors to the OpenDaylight Project. The goal of OpenDaylight was to create a community-led and industry-supported open-source platform to accelerate the adoption and innovation of Software Defined Networking or SDNs. Topics include:
Craig Corbin:
Hello, everyone, and welcome to another edition of the Broadband Bunch. I’m Craig Corbin, thanks so much for joining us. Software-defined networking, or SDN, is a dynamic, manageable, cost-effective, and adaptable framework, ideal for the high-bandwidth dynamic nature of today’s applications.
Craig Corbin:
It’s an architecture that enables network control to become directly programmable. And OpenDaylight, or ODL, is designed to deliver the benefits of SDN to support a wide range of use cases, including ones that haven’t even yet been invented. Software-defined networking, OpenDaylight, two of our topics today, and to help us on that journey, a man very much at the forefront of SDN and ODL development.
Craig Corbin:
Our guest has, for the past 14 years, served as chief technology officer for PANTHEON.tech, and is one of the three architects of the model-driven service abstraction layer. It is a pleasure to introduce and welcome the CTO of PANTHEON.tech, Robert Varga. Robert, welcome to The Broadband Bunch.
Robert Varga:
Oh hi, guys. Thanks for having me.
Craig Corbin:
It is absolutely a pleasure to have you on the podcast. And so much to talk about, exciting concepts, SDN and OpenDaylight. Before we even begin thinking about that, we need to learn about Robert Varga. Give us your background and tell us about who you are.
Robert Varga:
So, essentially I am a technologist at heart. I’ve been sitting behind a computer for the better part of 30 years and dabbling into all the things that come along. So, I essentially started as a developer for a company that actually developed software for visualizing weather.
Robert Varga:
So that was the first challenge and I learned a lot there. Then I kind of got bored of developing and having all these deadlines, so I thought I’d move into systems administration, which kind of was a job for five years or so. After which, I decided that’s boring, so let’s do some development again.
Robert Varga:
And a neat project came along, which kind of jogged me out of my seat and brought some very interesting use cases to the table and a very challenging environment. So, that’s how I kind of got into being the CTO of PANTHEON. And I drove the company in engagement there, along the lines or other during that time, we essentially discovered that there are other things that you can do with a computer. And that networks are not really what you see when you type in the URL or when you are a CIS admin and configuring IP addresses is all that you need, but there is a world of complexity and cool stuff, interesting stuff that goes on, on the internet and essentially to give you all that cool services that you get on your desktop.
Craig Corbin:
Absolutely. That may be the understatement of the day, the world of complexity because it is a tremendously complex world that we will begin to unravel, and greatly appreciate your expertise in helping us do so. And I guess to begin that conversation, let’s help those who are not aware of what SDN is. Talk about software-defined networking, because it is such a vital part of your efforts.
Robert Varga:
So SDN kind of got off way back in 2009, 2010, or thereabout, that’s where I first brushed with it. And the word of the day was OpenFlow and all these wide book switches and let’s tear down all that infrastructure we have and OpenFlow is SDN. But actually, we were working with major vendors in the networking space.
Robert Varga:
And for example, service providers were kind of iffy that, well, we are building this distributed thing called the internet, and there’s no way we can centralize it the way OpenFlow wants to do that. So, there was a disconnect because, at the end of the day, SDN is many things to many people, but the core idea is to automate networks as much as possible so that you don’t need to do have processes driven by people just to make changes to the network so that the network can adapt to whatever the reality is.
Robert Varga:
So if your users are coming online and they’re starting to stream more video then your network ends up optimizing itself to enable a more seamless experience. So, that’s-
Craig Corbin:
Beg your pardon, Robert, you touched on something that I think is important there with regard to the impact on providers and the fact that you talk about the network control is directly programmable. And that’s something that obviously makes an impact on the day-to-day operation for providers. Talk about that, if you would.
Robert Varga:
And so some of that is the entire DevOps thing that we’ve been seeing. So back when I was a CIS admin, which was something like 15 years ago, the way you would manage your network, your network administrator would essentially be sitting behind CLI and typing all those weird commands into their routers and all the firewall rules, all the tunnels and what goes on to actually provide a service had to be done manually.
Robert Varga:
So obviously that was a process that was prone to errors because people make mistakes and it was slow because well, typing that in into CLI takes some time. Whereas if you expose the same capabilities through a programmatic interface, you can hook programs, which do the same thing but do it much faster.
Robert Varga:
So your turnaround is faster and they can be tested. So they do how they’re written, right? So they don’t typically make mistakes unless there are bugs but you can fix bugs and they [inaudible 00:08:02] and don’t repeat. And so overall you can have faster turn-up of services and more reliability in your network.
Craig Corbin:
And that absolutely impacts service to the end-user. And there’s something that’s a component of this that I think is key that I’m intrigued with. And that’s one of the features of SDN is that it’s programmatically configured. And that means, allowing network managers to optimize their resources quickly. And actually, they can write themselves because the programs are not dependent upon proprietary software. Am I correct with that?
Robert Varga:
Well, there are all sorts of solutions. So, there are definitely locked-in solutions. You can buy commercial solutions from a vendor, but there is a large number of open solutions. And essentially the intent or other, the desire of most customers is not to get locked in with a particular vendor.
Robert Varga:
So they are more often than not looking for open solutions that they can deploy and then can take ownership of, and be confident that well, even a vendor failing or vendor jacking up the prices is not going to impact them as much if they were just locked in to a particular solution with a single provider and essentially migrating off of that would mean redeploying the entire solution again.
Craig Corbin:
And what you just touched on in part sort of segues nicely into the concept of OpenDaylight, when you referred to the fact that, if it’s implemented, if SDN is implemented through open standards and vendor-neutral, then we began talking about the impact of OpenDaylight. And again, a topic that many of our listeners perhaps are not familiar with in its entirety, helps us understand how OpenDaylight came to be.
Robert Varga:
So, the story, I don’t have the entire business background of where the idea came from, but essentially OpenDaylight was launched as a Linux Foundation collaborative project funded by the likes of Cisco, Brocade, IBM, HP was involved. There was a bunch of companies who realized that there is value in building a common platform on which we can collaborate and define our use cases and make them work, and then have commercial differentiators between different vendors who essentially will give OpenDaylight, tailored maybe with some neat things added on top of it.
Robert Varga:
So back then, that was 2013, all the early deployable solutions were commercial and they were anything but cheap, and you couldn’t really test drive them. The market leaders were I think QwikSwitch and [inaudible 00:11:47] and those sort of things. And they didn’t really cover everything that was considered SDN, right? So it was mostly campus networking, tightly controlled, a small number of switches, centralized control, and all these sorts of deployments.
Robert Varga:
And they didn’t do anything for service providers, most notably, and also things like broadband and different technologies. So the assumption was that you are going to do a Greenfield deployment for everything, you will buy new hardware to support these new automated use cases. Whereas as it turned out, open collaboration showed that you can do all these automated use cases with different technologies, and you don’t have to replace all the hardware. You can do brownfield deployments and kind of retrofit these new concepts and modes of operation into your existing network without a major investment.
Craig Corbin:
Absolutely. Well, let’s talk about that a little bit. And you touched on the impact on the broadband industry and with respect to just one use case being automation of service delivery. So that if you’ve got a broadband provider, they can perhaps include their scheduling with dynamic VPN services. And you’ve also got the benefit of both cloud infrastructure involved in that delivery and implementing the network function virtualization, talk about that.
Robert Varga:
Yes. And I’m really not into deployments these days. So, I kind of have to go back and kind of think about it a bit more while I do that, I’ll just show you the technology side of things where, well, we had these open flow things and mode of operation by provisioning and actually a bunch of program people came about and said, well, we do have this old protocol that nobody used and it supported on our CPEs.
Robert Varga:
I don’t remember the protocol name. And we actually can do most of what OpenFlow does through this protocol. And there is a plugin to OpenDaylight, which does exactly the same thing. So for use cases, essentially, this is where things get into hybrids, and you have so many options of how you can achieve provisioning that you essentially need to start with taking stock of where you are, what is the deployed base that you have, what are your CPEs, what are their capabilities, where are your users and all that sort of thing.
Robert Varga:
And kind of going to try to think of what is it that you want to achieve. And then you can go into the how, because essentially Kubernetes and all these virtualization containers and all these stacks that really matured over the course of the past five years have really changed the game. These days, you can essentially have any service delivered to you as a container. And all you have to take care of is bridging the gap between your customer’s CPE and wherever in your private or public cloud, you have that network function instantiated, and that can be either a VPN endpoint or any number of things, right? It can be a streaming service, it can be anything that your customer is accessing.
Craig Corbin:
And what you just mentioned is key. And I apologize for jumping in, but you just touched on something that truly is vital with regard to the impact on service providers. And that’s the ability to alter the time-consuming method that they are currently using with legacy approaches, with something that provides the ultimate in being nimble and agile in responding to customer needs. And that’s obviously a huge, important thing. From your perspective, where do you see these taking operations for an average service provider as they move forward?
Robert Varga:
So I think this will drive some of those. So, at first, it will be an introspection into your operations and how your say, back-office interacts with your operations. So when you have an ask for circuit bring up or for a service, then obviously this process needs to be automated and all the approvals and whatnot.
Robert Varga:
So, the first impact is getting really hold of your processes. And then probably the parallel track to do that usually is that you also need to understand what are your SLAs that you’re looking for? How does the network look like? How good are your network monitoring and processes dealing with faults? Because well, let’s face it, this is real-world and faults will occur no matter what.
Robert Varga:
And essentially, at the end of the day, the goal is essentially, maximize the capacity you can use in your network, so the raw capacity while also maintaining your SLAs.
Craig Corbin:
And that is such a huge part of the commitment that providers have to their clients, their customers are maintaining the SLAs, by the way, you are listening to the Broadband Bunch, our guests today, PANTHEON.tech CTO, Robert Varga, and Robert, so much to begin to unfold with these topics and concepts but want to change direction, just ever so slightly here for a moment.
Craig Corbin:
In the introduction, I mentioned that you’re one of three architects for MD Sal, the model-driven service abstraction layer. And that’s sort of your OpenDaylight controller. And if you would, give us a bit of an overview of that.
Robert Varga:
So the model-driven service abstraction layer, which is a mouthful, but really describes what it is, comes from the realization that your network management software and your network elements are deployed at different times. So in your classic deployment scenario, you would have an NMS and you would have a bunch of devices that are supported by that NMS.
Robert Varga:
Now, if you were to buy a new device before you can really manage it, your NMS needs to be updated to support it. And that kind of creates a network-wide hotspot, which is your NMS, which is the gating factor for upgrading your hardware. So, the key idea of MD Sal is that, well, it is centered first of all, around discoverability of network element capabilities, and this is coming from the NetCom protocol where the device can tell you what sort of configuration it supports and what is it that can be configured on it.
Robert Varga:
And based on that information, we’ve built tooling which can do this at runtime. So in the end, that network management piece of software doesn’t have to be updated to support new devices, because it is just a pass-through function, right? So it just mediates excess to it and the applications that you are actually writing to manage those devices.
Robert Varga:
So it can dynamically learn, well, this device supports these configurations, and it can project it to your developers or DevOps, really without the need for you to upgrade the centerpiece. So, it upgrades itself based on what it discovers in the network.
Craig Corbin:
So, that’s a huge savings time-wise, investment-wise. And I know that MD Sal when we look at the impact on providers in the broadband industry, it provides infrastructure services that allow data storage, service routing, the ability to publish services. Those are things that providers deal with every single day and are a huge responsibility. Talk about that.
Robert Varga:
So, we do provide a number of services there, which are really developer-centric. So you need to take them as a toolbox, right? So they’re providing you with a toolkit to actually realize the services that you need. And as you mentioned, there are huge savings to be had as you don’t have to do the same thing over and over again on different layers, but you can actually focus on what is the problem you need to solve, for example, device provisioning, right?
Robert Varga:
So, if you have a device coming up, then there’s a provisioning workflow that talks to the device and sees it as a device. Whereas, if you take a look at the same device from the perspective of a service configuration, it again has a different set of requirements and it talks to the device in a slightly different way. And most of the time, it talks to multiple devices to actually do end-to-end provisioning steps.
Robert Varga:
So that, again, the MD Sal here allows you to define the common set of concepts that are common for both of those views, and then use them freely since you are using them, that means that mapping the services to devise configuration is easier because you don’t have to write it. You’re essentially just copy-pasting the same template where copy-pasting is a turn of phrase, you don’t literally copy-paste it.
Craig Corbin:
Exactly. Well, there is so much to get into with this, and obviously, we won’t be able to cover it all in this episode of the Broadband Bunch, but I did want to get back to OpenDaylight and the project that you mentioned that sort of was the collaborative open source project hosted by the Linux Foundation.
Craig Corbin:
And, part of telling the story about OpenDaylight for those that are not as familiar is sort of looking at where we came from, where we’re going. And I’m always curious about what kind of reception in the industry that gets because I know that one of the descriptors can be that we’ve gone from a turnkey approach to a toolkit approach for providers. If you would, give your thoughts on that.
Robert Varga:
Exactly. And that has been the evolution. So, when we started OpenDaylight, the idea was to build really a network management system where you would deploy a huge server with a huge amount of memory, and it would manage all of your networks and it would be a turnkey solution, you would just deploy and it would just work.
Robert Varga:
As it turned out that A, is a very ambitious goal. And we quickly learned that the use cases and the deployment details for essentially every provider be it small, be it broadband, be it fiber, whatever, is very different. And to the point where you really don’t want to have this big box which does everything, but rather you want to have a toolbox of useful tools, each of which does one thing and does it well, and when you combine those tools to essentially create a bespoke solution for a particular deployment, which takes into account all the requirements and all the constraints, which essentially boils down to, well, you cannot well, really expect to have a turn-key solution for every deployment.
Robert Varga:
And there may be some bits and pieces missing. So when you’re creating such a deployment, you actually fill in those bits and pieces, and then decide whether they’re really specific for this deployment, they’re not reusable, or is this something that is reusable, and then you essentially put it back to the toolbox, which is OpenDaylight.
Craig Corbin:
And, in part, I guess what you’re doing there is providing the ultimate in self-control on the part of providers and their teams of developers in managing their operations in a much more concise and agile way, which is sort of a phenomenal concept. And I’m curious, not only with what has been the reception of OpenDaylight to this point, but I would love to get your thoughts on where you see OpenDaylight as we move forward, the future of the OpenDaylight project.
Robert Varga:
So, OpenDaylight as such is maturing its bits and pieces. And it is a critical part of, for example, the over end-use cases, so the open the radio network and 5G use cases, and while it doesn’t provide the end-to-end deployable for a service provider, like for example, ONAP would, it provides the critical infrastructure bits and pieces to enable that deployment.
Robert Varga:
So going forward, what we are focusing on is really hardening those bits and pieces and really prepackaging individual use cases so that you can leverage pre-built containers to build your network control functions if you will. So rather than having a single container, which does everything, provide a specific blue container to talk to a specific class of devices or perform a certain or a management function, and then have those assembled to form a complete solution.
Craig Corbin:
And you touched on containers and containerization. And am I correct in that approach provides the ultimate in flexibility for providers to very quickly alter their services and their operations, whereas, in past ways of doing things, it would be a very lengthy process, is that correct?
Robert Varga:
Yes, because essentially your container is a small unit of integration, right? It has a well-defined API, which is not all-encompassing. So, you know the interface, you can test it quite easily. It is an isolated bit. So if you replace it and something goes wrong, you can really quickly go back and replace it back without having to replace everything in its surroundings. So from that perspective, mixing and matching these containers and these functions allows you to evolve the solution in a much more agile way.
Craig Corbin:
And that to me is as intriguing a concept as anything because of how it allows providers to make changes without impacting other parts of the operation and to do so in a matter of seconds and minutes, as opposed to days with legacy approaches. As we begin to wrap up this visit, I’m always intrigued also by what drives our guests here on the Broadband Bunch. And it’s obvious Robert, that you have such a passion for the work that you’re involved in. And knowing that its potential impact on communications technology is immense. What drives you? What’s the driving passion for Robert Varga?
Robert Varga:
Oh, it’s always learning things. Learning things, improving things, making things work better than they used to, because essentially if I have a month when I don’t learn anything new, it is sort of like, well, nothing interesting happens. So, learning and improving things around.
Craig Corbin:
And you’re doing such a phenomenal job of that and the last question to you, obviously, with projects of this nature, there is so much that transpires from day to day, year to year, and many times you come across developments that you wish, perhaps that you had known about much earlier. If you could jump into the time machine, go back several years, is there anything that you would tell yourself at an earlier time in your career about taking a shortcut?
Robert Varga:
Well, there’s just so much, and knowing myself, my younger self probably wouldn’t listen. Probably it would revolve around keeping things simple and easy to understand, rather than trying to build nuclear submarines would have saved quite a bunch of effort, I think.
Craig Corbin:
I love it. Robert, it has been an absolute pleasure to chat with you today. So much to unfold with these concepts. We greatly appreciate your expertise in helping us begin that journey. I would love to have you back for a future visit to again, dive a little bit more deeply into certain aspects of this, but thank you so much for your time and expertise with us here today.
Robert Varga:
Thanks for having me.
Craig Corbin:
Absolutely. That’s going to wrap up this edition. We thank everyone for allowing us to be a part of your day. We’ll see you next time right here on the Broadband Bunch. So long, everyone.
Join us here on the web at Broadband Bunch, to see the latest episodes, news, and information. The Broadband Bunch is sponsored by Utopia Fiber, DxTel and ETI Software.
