SD-WAN began life just a few years ago as a means for enterprises to cut costs on expensive MPLS connections, but it's matured far beyond that. Enterprises are looking to SD-WAN for financial benefits, but also to provide application delivery -- particularly important as applications move to the cloud -- as well as insights and reliability. And MPLS isn't going anywhere.
To help enterprises migrate to SD-WAN, we produced this short podcast, sponsored by Citrix Systems Inc. (Nasdaq: CTXS), featuring Sterling Perrin, a principal analyst with Heavy Reading , and Carl Brown, Citrix senior director of product management for NetScaler. The podcast is moderated by me, Mitch Wagner. Listen to the podcast by clicking the play button below, and read the transcript -- lightly edited for readability.
Mitch Wagner, Editor, Enterprise Cloud, Light Reading: Let's just jump right in. SD-WAN has gotten a lot of hype in enterprise networking. Sterling, why don't you give us an overview of SD-WAN?
Sterling Perrin, principal analyst, Heavy Reading: Plenty of hype, but quite a bit of reality beyond that. At Heavy Reading, we've been looking at it for a while now, and we do have a definition, which is essentially a software-defined overlay approach to connecting, controlling and monitoring enterprise wide area networks, or WANs. SD-WAN integrates a centralized controller function, essentially. It relies very much on software automation, so it's very much within the concept of SDN, applying it to the WAN.
The drivers are pretty significant here. It's really about the enterprise migration to cloud-based applications, which has presented a special challenge for enterprise branch locations. Historically these things have relied on low-speed T1s, typically MPLS over T1 hubbed into corporate locations from all these branches. It was very good historically as an architecture or network type, but as the world moved to cloud it became a very poor way to do a lot of those cloud-based applications. That became the main driver for the advent of SD-WAN. So, very much around cloud and branch.
Wagner: Great. Carl, anything you want to add here?
Carl Brown, senior director of product management of NetScaler SD-WAN at Citrix Systems: There's not much I can add. Sterling did a great job with a very comprehensive definition. I'll just say it's in line with how Citrix defines SD-WAN. It's in line with how a lot of others in the space do. It generally resonates with what we're hearing from our customers. Yeah. Great job, Sterling.
Wagner: So, it sounds like SD-WAN is about a lot more than lowering costs. How have things changed over time? Sterling?
Perrin: If you look back at the advent of SD-WAN, and really it hasn't been that long, this is a trend that just kind of skyrocketed over the past two years, maybe, of getting it to mainstream attention. Originally, it was viewed as very much as a replacement for MPLS. I mentioned MPLS when I was going through the definition there. That was maybe the panacea of what SD-WAN would do. What happened is enterprises found that you can't really replace all of your MPLS. There are things that MPLS was doing poorly in the context of the cloud. The amount of bandwidth was too low and the price was too high, and this hub-and-spoke architecture was poorly suited. That was the downside. But there's some really good things that MPLS does in terms of reliability and manageability for the network.
What we've seen is SD-WAN is more of an enabler for this concept of a hybrid WAN, which includes multiple access options from the branch, so you continue to have MPLS, but you have DLS, cable, even LTE, Ethernet. Lots of different options, and this software-based programmability of SD-WAN is a way to really define policy and then move between those different access options.
What that means for MPLS specifically is the enterprises are not necessarily getting rid of MPLS. They are capping their investments, and they're putting a lot of the new stuff on the SD-WAN, but there are still a lot of reasons why you need MPLS in the network. It's really around this hybrid WAN. Yes, you do reduce costs, but it's not like you've thrown out all of your MPLS and reinvented how you do networking completely.
Carl, I don't know if you've got anything you wanted to add on that.
Brown: Yeah. I think I like what you said. I would just press the point again on reliability. A single WAN circuit going into a branch location puts you at risk for performance issues or blackouts where the line goes completely. Reliability is something that our customers tend to highlight as one of their drivers, as opposed to just simply reducing the costs for MPLS services.
That said, all enterprises are different, and their drivers vary. But going back to your definition, it's part of moving the applications to the cloud. And that's another driver that I see [as] really common, where an enterprise has made some decisions around migrating from a legacy app to a SaaS based application, or they've refactored an application and have it running in a public cloud environment. They want to re-architect the wide area network to provide better and more efficient and reliable access to where these applications are now being delivered from. That brings up the hybrid WAN architecture, and soon as you start heading down that path then SD-WAN really makes sense.
Perrin: Yeah, I agree completely. The hybrid WAN really has been the enabler, I think. One's enabled the other. SD-WAN's enabled hybrid WAN. Hybrid WAN has given the main use case for SD-WAN, for sure. And I agree completely with your points on reliability. When we ask enterprises -- as well as when we ask service providers who are supplying the enterprises -- that reliability comes up at the top of the list in terms of requirements for the networking.
Wagner: SD-WAN has expanded its focus over time. It's now focused on application delivery, insights and reliability, but it's still got to provide cost benefits, right? So if that's the case, do the carriers see it as a threat? Carl, why don't you take this one to begin?
Brown: Rather than framing it in terms of cost, let's just frame it more broadly in terms of financial benefits. Reliability has lots of financial benefits. If the network's down, a business can't transact. If they're waiting on downloads or waiting on print jobs, they're unable to respond to customer's inquires because the network's slow. That has financial impacts. SD-WAN can prioritize applications that are mission critical. It can make those applications more reliable. It can make the applications work better, which has benefits to the employees. It has benefits to their customers, therefore improving productivity. It can be used to drive digital transformation by instrumenting more things, more processes, or more services within a branch office location.
Sure, there are some opportunities for cost reduction, which Sterling highlighted earlier, but there are more financial benefits as I mentioned. To some extent, change is a threat, and so controlling bandwidth costs, which is what enterprises want to do, therefore would be a threat to a carrier. But at the same time they can see it as an opportunity. Yes, they might be selling more bandwidth for a same price or a lower price, but it's also an opportunity to provide additional options or additional functionality to their customer base, because making this migration from an MPLS only based WAN to a hybrid WAN is complex and creates opportunities for the service provider to be a better player.
That's my perspective. Sterling, I know that you work a lot with carriers, and maybe you could give us their perspective.
Perrin: We've talked about things so far from the enterprise view. Of course, the enterprises are the main users of the service, but as you indicated the network operators have gotten into selling SD-WAN as a service, as a managed enterprise service the same way that in the past they've offered MPLS services, Ethernet services, and a whole host of other things. In terms of the threat, and we've done a lot of research in this area, in terms of the threat it does get back to that whole issue of the MPLS cannibalization of business. If the operator is selling SD-WAN in place of its own MPLS service, then clearly they're taking business from one side and moving it to another. If in this case SD-WAN was a lower cost, then they'd be losing money. In many cases, I think the way it's rolled out is SD-WAN is not less money. They keep the cost about the same.
But anyway, there's a trade-off there. I guess a couple of points I make on that from the many carriers we've talked to and surveyed, the operators have realized that SD-WAN is inevitable. You run into that situation as if they don't offer SD-WAN to their own customers, then someone else is going to come in and do it. They need to do it. It's just the way that the market has moved. It's the way enterprise demand has moved. One other point that's been pretty critical in how we've seen MPLS and SD-WAN move forward within network operators is it's actually very similar to the point I made about the enterprises themselves. They haven't completely thrown out the MPLS, it's been more cap MPLS as opposed to replace. This gets into that hybrid WAN use case, which network operators have actually doing tremendously well with.
I guess the last point I'd make on this is there is a very real issue for some operators that do have a very large installed base of MPLS that they're going to be hesitant to move strongly into SD-WAN because there's so much revenue on that MPLS side.
Putting the enterprise hat back on, I would say that the enterprises do need to be aware of this, and they do need to do their due diligence as they talk with network operators to understand where they fall within the spectrum of how much MPLS business [they have] and how they're really approaching [SD-WAN], because there is a threat there. I think the benefits of SD-WAN both to the enterprise and to the service provider outweigh the threat, but you still need to be aware of it, especially as you're an enterprise looking into this.
Wagner: Reliability is very important for SD-WAN in delivering applications, whether those applications are coming from the data center or from the cloud or software as a service (SaaS). Does that mean that relying on a single carrier is a problem? Does relying on a single carrier go against the concept of reliability? What's the right approach here? Sterling?
Perrin: Yeah, that's a big question and a good one. I guess the best way to start this off is that there's multiple models for how SD-WAN is purchased by an enterprise. I'll just tick through them, and then we can discuss them each. One of them, which we've talked quite a bit about, is the enterprise do-it-yourself approach, which is where the enterprise has full ownership of the SD-WAN product. It's generally software with a hardware component, maybe, to it as well as all those connectivity pieces and piecing those together. That's a pretty big time and expertise investment on the part of the enterprise.
And then on the far end of the spectrum, which we've just talked about a bit, was this network operator managed service where the operator is the one offering SD-WAN as a service to its enterprise customers. In this case, the operator's selling you the SD-WAN software and hardware pieces as well as all that connectivity and managing those relationships.
And then I think somewhere in the middle is this system integrator approach or a managed service provider approach, primarily driven by resellers. If you think of a reseller selling not just the SD-WAN software and the software pieces but also piecing together those network elements to it. I think those are the three main ones. Within the spectrum of that, you have different levels of how much you're relying on one particular service or one particular component of that overall SD-WAN piece.
Wagner: what I'm hearing is there are three different approaches for an enterprise to deploy SD-WAN. What's the right one?
Brown: Mitch, the right one varies from enterprise to enterprise. If the enterprise needs the ultimate in flexibility and has the right expertise in house and is used to operating their WAN today, DIY might make sense for them. If the enterprise is looking to move to SaaS and doesn't have the support resources internally to manage their wide are network or manage the complexity of the migration of their wide area network, then they might want to go with a network operated managed service that's delivered from a traditional telecom provider.
What we see emerging as becoming very, very popular is the in-between model where there's a system integrator approach delivered by a managed service provider. It sort of gives them benefits of both ends of the spectrum, the do-it-yourself or the network service provider model, but yet still gives them some measure of flexibility to put the right pieces together. So, that's popular emerging model. But again, I think it goes back to the right approach varies based upon the enterprise's needs.
Wagner: Let's wrap things up. Last words of advice for our listening audience? Carl, why don't you take this one?
Brown: Right. There's a lot of noise out in the space on SD-WAN, and what I always encourage customers is first define your objectives in transforming your wide area network, especially from an application perspective. Think about your needs today for applications. Are they real-time? Are they interactive? Where do you think you'll be three to five years from now? At the same time, look at what IT functions you want to handle in-house over that three to five year period that we spoke about vs. what are the functions that make sense to look to a service provider to provide to the enterprise.
Once you understand what your needs from an applications perspective and what you're prepared to commit to owning and managing, it helps you understand which model works -- do it yourself, with a managed service provider or network service provider, which model makes sense to you. Ultimately as an enterprise, you have a lot of flexibility, and you can define in many cases with a managed service provider or network service provider what technology you want to be used, what vendor's appliances you want to be used in the service that they deliver you around SD-WAN.
And think long term. The products are evolving, the technology's evolving, and who the right vendor is today might not be the right long-term partner because who's going to have the best solution three to five years from now? That's the time horizon you should be looking at.
Perrin: Carl's made some good points there. I guess the first thing I would say is just the very basic point, SD-WAN is very real. It's not a fad technology, and it's having a very huge impact on the market now and going forward, as Carl said, three to five years. This is the way to think about how SD-WAN will roll out in enterprise networks. Now is the time. If you're not on this trend, now is the time to certainly get into it and dig in.
I just echo another point he made, as well, about the number of choices. As we look at the vendor landscape, there's a lot. There are too many, to be honest. A lot of investment went into startups in this area. Enterprises need to choose their partners carefully. There's a lot of good technologies out there, but you also need to look at the company and how stable are they, what do their customers look like, do they have big customers to anchor them, keep them going forward.
And I guess the last point is just as I was thinking about when Carl was talking on the last question about the different spectrum, from do-it-yourself enterprise to the reseller-managed service approach, to the network operator -- I think part of what's driving this is we are seeing the market for SD-WAN move downmarket. The very early adopters were the big enterprises like Coca-Cola and Gap and huge, huge companies, huge banks. We are going to see this, as [with] other trends, move downmarket into mid-size enterprises. As that happens, then yeah, a lot of these companies are not going to be in the position of do it yourself, so the systems integrator approach or network operator approach, those are where there's going to be a tremendous amount of growth going forward. And I would just encourage the companies that fall into that kind of category of company -- or maybe they just don't want to take it on -- to know that those options are out there, and they're very viable options beyond doing everything yourself. It's the way these IT trends and telecom trends have always emerged over time. I think SD-WAN will evolve in a very similar fashion.
Wagner: Okay. Let's leave it there. Thanks for joining us, everybody. I'm Mitch Wagner from Light Reading and we've been talking with Sterling Perrin from Heavy Reading and Carl Brown from Citrix Systems.
— Mitch Wagner Editor, Enterprise Cloud, Light Reading