Well, hello everyone, I really delighted to welcome you to TNC's latest podcast on the topic of MPLS versus SD-WAN, versus SASE. I'm John Waterhouse, CEO of TNC, and I'll be your host for the next 20 minutes. As I'm sure everyone joining knows, TNC is the UK's largest independent network and telecoms strategy and sourcing consultancy. We support over 280 major UK multinational organisations, and help them to get the best possible commercial, technical, operational, and contractual results from all aspects of their network and telecom services.
So joining us from TNC today to share his insights is our CTO, Craig Northveth. Craig, do you want to say hello to our listeners and viewers?
Yeah. Hi, John. Hi, everyone. Welcome back, I'm looking forward to today's session.
Fantastic, and it is a cracker. I have to say this is probably my favourite topic of all - and so I'm really excited about discussing it with you. So yeah, comparing MPLS, SD-WAN, and SASE - so really, looking into this whole revolution, which is sweeping the enterprise network market, what could be better? So yeah, let's set a bit of context, and we'll get straight into it, Craig?
So yeah, you know, I mentioned this technological revolution that's sweeping the market; and this is something we've been tracking for a while now, a good two-plus years. You know, I think the driving force behind this revolution is, well there's a number of factors, right? So - workloads, shifting into the cloud, demand for enhanced network and application security, virtualised infrastructure, network flexibility, vendor technology disruption - but I think what we're finding is the pace of all of those changes, and the complexity of those drivers, and the need to get it right, is meaning many of those organisations are struggling to know how to move forwards, how to realise those benefits: 'What's real?' 'What's vapourware?' 'What's...?' you know, etc, etc. 'Is it time to move away from MPLS?' 'Does MPLS retain a role for the future?' 'Is SD-WAN ready for primetime?' 'What even is SASE?' 'And, you know, is that a realistic deployable solution yet?'
So you know, I think in summary, we see a lot of excitement about the possibilities of these solutions, but a lot of nervousness about how to go on the journey. And of course, how to ensure you reach your your target destination, but never fear, because, on today's podcast, we're going to break it all down for you and talk through exactly that journey. And I think in particular, we're going to focus on the four key transitional states, which you've been defining Craig, you know, and particularly obviously, that's going to define the likely start point for most organisations. So let's get straight into that, let's look at our first topic: so Craig, do you want to talk us through what those four transitional states are? And you know, where most organisations are likely to be today?
Yeah, well, John, absolutely. I think you've touched upon a great point there. You know, we've been I guess what, 15 years in this business now, and I think if we look back over, maybe the first 10 of those years, there's limited selection in terms of availability of networks in the market, you know, it was kind of MPLS or nothing, should we say, maybe a bit before that, there could have been a few other choices but yeah, let's say, let's pretend we're not that old.
This will be our first podcast where we mention frame relay - you know...
It could be, it could be - I didn't want to go there , I didn't want to go there. So let's yeah, let's just pretend that the past is MPLS. And so yeah, I think if we start our journey there, in terms of the transition you know, a lot, of our customers back through the years have had, kind of, MPLS only networks - and we still do have some customers that effectively opt for that model - you know typically, this is a closed network, it may have some sort of external gateways embedded into the network, or you may break out to sort of public internet services via the Data Centre. I think, you know, this type of environment typically will be provided by a single service provider, you'll have some contractual guarantees around levels of service, quality of service, but typically, it's quite a high cost solution, and it probably lacks a bit of flexibility in terms of, you know, being able to consume high bandwidth, or being able to get to cloud services. So we're starting to see some limitations around MPLS being the only model, or being the preferential model.
Okay, so if we look at Phase Two, what we're seeing in terms of the revolution is this model of hybrid networks. And so, this is quite simply introducing internet services into the mix at a local level. So this is preferential because you know, backhauling internet traffic or 'Software as a Service' type of traffic across the backhaul MPLS network and breaking it out centrally, is just not an optimal route to take that traffic. So we start to see the introduction of local connectivity, in line with the hybrid MPLS networks as well.
That kind of leads us on to, I guess, more more current day, and where we're seeing the introduction of SD-WAN networks. So we're going to classify this as sort of a programmable network, so this is where we're starting to see abstraction of the overlay and the underlay. So a programmable network, and particularly kind of when we talk about SD-WAN, we can still run SD-WAN over MPLS networks, and we still do see customers effectively having MPLS in the mix in these programmable networks, but more often than not, and particularly as we look at international networks, the preferential underlay service is now internet, and we're starting to see a lot more services being delivered as a SD-WAN overlay with an internet underlay service. The reason for this, I guess, is a couple of things. So the benefits of being able to have a selection of underlay providers is you've been able to kind of source your internet services, even regionally or even on a local basis, so there's cost efficiencies in doing that. The SD-WAN overlay obviously, because it's abstracted from the underlay, it gives you a centralised control plane. So that can be orchestrated from a policy perspective, you can embed security into there. So there's a lot of advantages moving to this model. And I think as we we go on in a second John, to talk about some of the factors around what phase you should opt for, we'll touch upon some of the benefits as well.
And just thinking about that Phase Three, talking about the abstraction of the of the underlay and the overlay, does that, I get that at a technology level, but does that likely, therefore also mean you've got a separate service provider? Are you buying your underlay from these service providers and the overlay from somebody else? Or is that the same service provider? What's the trend there?
So I think what this introduces is a level of choice. So you can if you wish still have this as a single service provider, we've seen a lot of the telcos have tried to productise SD-WAN services now and effectively embed them on their underlay services. But you can also break it down so that abstraction allows you depending on the geography of your environment, depending on how you want to kind of operate the model, operate the environment going forward, you can separate it out. So you know, for instance, in a global network, you could have a single provider providing the management of that overlay environment - the SD-WAN infrastructure and the services supporting that - then you could have multiple providers that underlay again, depending on depending on the geographies, depending on the density of sites and locations, depending on the number of circuits you've got. Obviously, there's a tipping point around where that comes both sort of commercially feasible, but also operationally feasible to do that. But we do see both routes being taken.
And presumably therefore what you're getting there is, you know, is that eternal trade-off we talk about, of complexity versus cost, versus performance, you know, so if you have multiple underlay providers, potentially that's cheaper, you can look at, you know, lots of in-country providers, but it's complex to manage and orchestrate. And yes, yeah, okay... I know, you're keen to get onto the the sort of Phase Four and that potential sort of end stage of the journey, but just going back to sort of what's driving customers out of Phase One into Two, and onwards into Three, you talked about some factors. We've talked before about a mix of sort of, what one might call push factors and pull factors. So the pull factors being: positive benefits those organisations are seeking like, lower cost or greater flexibility, or more performance, access to the cloud, or whatever it is, presumably, we're also seeing some push factors in their dissatisfaction with telco performance and that sort of thing. Is that a factor?
Yeah, I think it is, you know, we're starting to see increasing demand from two angles really. So we're seeing, dependent again, dependent on the type of organisation, we're seeing increasing demand either from the business to push more applications, push more services, be quicker to the market with certain things, and that requires a level of agility in the network which, when we start looking at the kind of the MPLS, and even hybrid networks to a certain extent, there's a limited level of flexibility in their networks to react quickly to things like bandwidth changes, or security changes, or policy changes, or performance-related changes. It's quite difficult to implement that quickly in those models, which is why organisations are starting to look at something that's far more agile, that you can relatively easily flex bandwidth, but by adding additional services, you can manage performance more centrally, you can get better visibility of what's happening, you can secure traffic more easily. So it's around the kind of ease of management of the network where we're starting to see real drivers or organisations to deliver against the business objectives and demands, and seeing that as a benefit.
I think the other side is customers as well. So again, you know, customers now, and this drives, I guess, behaviour from an application point of view, from a business point of view, they want more efficiency, they want to be able to consume services, you know, from from mobile devices, from apps. So again, that means more of an agile network environment to allow flexing up and flexing down and services more in line with with demand. So every business demand or customer demand is just that level of agility that's required in the network now, which, you know, people are starting to look away from the more traditional MPLS hybrid networks into a more dynamic, programmable network space, and, you know, in the future into more intelligent networks as well.
Yeah, it is really interesting and look, I'm not going to delay you from getting to that sort of Phase Four. But you know, I guess, you can really sort of see how the terminology has changed, that when these sort of intelligent networks, and some programmable networks, and ultimately intelligent networks were first starting to be kicked around and talked about, it was very much discussed as a cost saving objective, but really we're increasingly saying, - look at your cost you know, well of course, it's a factor but actually, you can't operate your business with a Phase One network anymore, you've got to be looking into Phase Three, and potentially even Phase Four, to meet the demands that have been put on you by your business stakeholders.
Yeah, and I think, you know, in recent times as well, the pandemic has caused a slight different rethink around using things like SD-WAN. So, you know, in that kind of programmable network space, we're obviously starting to see quite a big uptake in SD-WAN. So a lot of the service providers have productised that now, it's pretty much the default mode of operation for the RFP process. And, you know, we've seen that pretty much default, commonplace as a proposal. The problem, I think that the pandemic's introduced is that, SD-WAN's great if you're in the environment, if you're in the branch if you are sitting behind the appliance - so you know, it provides all that performance, and security, and application, and control, and dynamic networking effectively - the problem is right now, and I think what we're seeing going forward is that we're more than likely to return to some kind of hybrid working environment. So therefore, that SD-WAN infrastructure, that is sat in the branch or in the remote sites, is not going to be as useful as it once may have thought to have been, back before the pandemic, when everybody was in the office. So that's kind of where the the thinking now is moving on to this more intelligent network model where you know, SD-WAN probably sits within the intelligent network model, but you've also got the capability of accessing some of these security services, and policy and performance services, whilst you're off the network - at home, for instance. So this is a model where we're starting to see - the term you referred to at the start - SASE, the Secure Access Service Edge? This is a model where we're starting to see some of those controls that traditionally would be sat on the edge of the network, actually be elevated up into the cloud, or the edge of the service provider network. So your last mile between the device and the edge of the network effectively, is commodity type services. But then you're consuming the security, the routing, the policy, the application performance, at the edge of the network. So that's available for everybody, whether they're in the branch, or whether they're working from home, or from Costa, or McDonald's, or wherever, they're getting the same experience, and they will be getting that same level of security and performance, whilst from any location. So this is where we're starting to see this, this concept of SASE come in.
I think beyond that, and looking at where the roadmap for this kind of environment is going, it's becoming much more, almost like Pay As You Go type models as well, or 'Network as a Service' model, if you like. So all this stuff is effectively virtualised at the edge of the cloud network, or the telco network. It's moving into similar model of 'Cloud Compute', where you spin up services, you create functions, and you create paths and bandwidths for when it's required, but then you tear it back down again, when it's not being used. So it's a constant, - it's almost like an infinite resource that can be reused all the time, but it's not nailed up all the time so you only use it when you need it. And then you start moving into this more sort of consumption-based networking.
And presumably one of the things that underpins what we've talked a lot about, both on podcasts, and videos, and breakfasts, and just the two of us as well, about what the operating model is going to end up being for all of this. You know, I suppose one of the challenges with what you're describing is the traditional model of - human being logs into router, starts coding, updating configs - that's not going to persist into this kind of intelligent network environment, right? Because t can't be dynamic enough, the resource overhead is too great. So presumably, what you're also talking about there is starting to get into a level of automation and orchestration to support those those objectives?
Yeah, so it kind of becomes an intent-based network then. So the intent drives the automation, which drives the creation of these functions, it drives the creation of the paths, it drives the creation of the policies, and the security applied to those paths, based on the intent of the traffic if you like, or the user. So, if the user is trying to consume a certain service, that intent will drive the automation to build that service, to instantly almost, create that service. And, once that service has been consumed, tear it back down again so it's really driving a high level of automation, which, you know, is constantly learning as well. So it's using AI, it's using machine learning, it's constantly kind of making itself more efficient, more cost effective. But yeah, it does, it does kind of ask a big question around - well, in the future, what what role do telcos play? What role do service integrators play? What role do, just general, internal operations teams play in in developing this kind of stuff? Because it is changing rapidly to meet the current demands of what we're seeing across the marketplace at the minute.
Okay, so that's really interesting, Craig. So I can see sort of how those four phases progress through from the sort of start point for most organisations, with a sort of fairly traditional looking, probably MPLS network, moving through hybrid into, programmable, and ultimately into a sort of intelligent network. And is it that simple? Will an organisation start at Phase One? And, step into Phase Two? And then Phase Three? And Phase Four? Will everyone end up at Phase Four? Might they you know, and is that pathway that linear?
It's not linear, no. I think if we look a few years back, from Phase One to Phase Two was fairly linear, most organisations went from those kind of closed MPLS networks to hybrid networks. I think just based on the points that I've made there now around some of these indirect factors like COVID, for instance, and the way that people are now working, if we think about a future where we're in hybrid model, then actually, is there an opportunity to sort of almost leapfrog the programmable network phase, and go straight to intelligent networks, move more into that SASE / Zero Trust network architecture model, where actually you're consuming 'Network as a Service' rather than building? You know, it's modern, but it's still traditional infrastructure at a branch model? Do you just move away from that completely, and elevate everything into the cloud and consume it as a service? I think we'll start seeing more of that in the coming years.
So it would it be fair to view these more as potential end states, rather than stopping off points on a journey?
Yeah, certainly Phase Four intelligent networks. I think it's a big step let's say, for a closed MPLS network to go 'Let's move everything to the cloud, and let's go Zero Trust, and yeah, let's just do it overnight'. That's not going to happen. But it could be strategic transitional phases that get you to that point, and whether you go through, hybrid to programmable, to an intelligent network; or whether you go straight to programmable, and then to intelligent network, or kind of a mix of the two together? That's probably where we'll see most people end up, where they've got some integration, or at least, some co-working between the programmable intelligent network models. We will probably see most people end up around that space.
That's very interesting. And what do you see as the strategic outlook? So if you if you're looking two to three years out, how do you see this changing?
I think there are two things, and we've touched upon on the COVID piece. So you know, the external factors that will drive a change in network strategy will be things like, yeah, we're not going to be having people in the office anymore , (or we're not going to be having as many people in the office anymore), that's obviously going to be quite a significant impact to the network that you provision within each of your locations. So you know, maybe we'll start seeing more organisations moving away from that sort of closed, secure private network to maybe just an internet based network in the branch location so they can have a more of a frictionless experience between the home environment and the work environment - opening it up, but then consuming everything from the cloud. So I think we'll see a bit of that, and we're already starting to see a lot of organisations considering what their strategies are post-COVID, and what ways of working will change in terms of when they're in the office, they might be hot desking or choosing meeting locations. So we've seen a lot of things there, which will be direct inputs or requirements into future network strategies.
I think the second piece really then is around this concept of, almost like utility-based networking, or 'Network as a Service'. So starting to see the last mile services and even some of the edge infrastructures become slightly more commoditised and a lot of the intelligence moving out closer to the core, or close to the cloud networks. So whether that's services that we start seeing through the big cloud vendors (like Amazon, or Microsoft, or Google), potentially offering some of those services, I wouldn't be surprised at that. But we'll also see the telcos trying to do a bit of a fightback I think, because, they are obviously concerned that they can see this coming, that a lot of their wires-only services are going to become commodity services, and we already know, there's not a great deal of money in them, so they've got the ideal infrastructures to be able to provide this virtualised edge. So you know, I wouldn't be surprised if we start seeing a lot more of that coming down their roadmaps and then starting to see productisation of SASE services from them as well, and so I think that's the strategic outlook.
Some of the other things, and just slightly going back to some of the Phase progression, there are obviously some direct factors as well - some obvious things like compelling events, like contract end dates. So you know, contracts ending on MPLS, do we want to sign another five year MPLS contract? Probably not. I think a lot of people will be quite nervous doing that right now. Because you know, it's a bit of a lock in, and you obviously need that level of agility and flexibility in the kind of unknown world that we're living in. So, yeah, I think that's a key thing that also challenges as well.
Of course, always. And presumably, there's an element, as you say, with, you've got an MPLS contract coming to an end, and just because you picked a service provider five years ago, because of their great MPLS solution, there's no guarantee that they're going to be the right people to take you on the next journey. You know, do you want a single supplier to do that? Might you want to break it up with the underlay/overlay model we talked about earlier? But you know, other service providers are going to be, you know, perhaps more progressed with their programmable intelligent network capabilities, and so on. So presumably, that's also driving quite a bit of change?
It is, and it's back to one of the first points we made, customers have got a choice now. I think, back 5-10 years ago, maybe not even that, maybe less than that, the choice was fairly limited in terms of technology and therefore, in terms of suppliers as well. They've got a choice now though, on how they want to build or architect their networks. If they do want to do a slightly different operating model, if they do want a higher level of agility or co management, if they do want some more control over certain aspects, they have that choice. Now technology has allowed them that choice. So they are thinking slightly differently around how do we how do we meet our requirements? How do we get that agility? How do we meet our business demand? If we provide flexibility for business demand changes, then how do we ensure it's secure? All of these things, there's a lot more choice and a lot more options available to deliver against the requirements. Rather than just saying, right it's an MPLS network, so we'll go to one of five providers or something.
Yeah, absolutely. No really, really interesting stuff. Sadly, as always with these podcasts, they're fascinating, and then, I realise we're basically running out of time. But, you know, just in terms of very quickly, what we would recommend that customers do next with this, I think, probably the single biggest takeaway we would give any organisation trying to decide on what its future network journey is, - give us a call. Let's talk about a really detailed strategy development process, because probably, doing the same things over and over again, is almost certainly not a feasible outcome. But navigating this very complex landscape - yes, now you've got lots of choice, but you've also therefore got a lot of opportunity to make suboptimal choices let's say, and obviously that creates a lot of risk given the criticality of the network to your IT infrastructure. So I guess that would be the key takeaway right? It's that you've really got to lean into that strategy development process?
It absolutely is. I don't know if we have said this on the previous podcast, but you know, a network strategy was developed a year ago, is highly unlikely to be relevant today, just because of everything that's changed from a technology point of view, from a world point of view. Almost everything has changed. So yeah, it's looking at revisiting that, ensuring that you've got the requirements again from a top down point of view - what the business is trying to do, what you're looking at, from a states and facilities point of view, how are our people going to be working in the future? What's the best way to consume services?
Taking all of these sort of inputs, and defining, and aligning them to the choices that are available in the market and trying to build a roadmap that kind of paves the way to success I guess, is what we do for lots of our customers, so we've got lots of insight and expertise in that space, but yeah, I would definitely recommend people think about that, and consider if what they have got written down today is still relevant.
Craig, fantastic as always. Always fascinating, always interesting but sadly, we're going to have to draw it to a close, our 20 minutes is up. But always grateful for your very interesting insights. And, thank you everyone, for listening and watching. Please do let us know any questions that you have about this, or any other network and telecoms topic. You know, we love talking about it. You can get in touch through our website - networkcollective.co.uk, or any of the usual social channels. And we look forward to talking with you again on our next podcast.
Thank you, John. Thank you, everybody.