IP VPN vs. Ethernet services
Date: Wed, 09/07/2011 - 19:44
At the beginning of the millennium, managed IP VPNs were rapidly replacing costly and inflexible legacy WAN alternatives such as ATM and Frame, primarily as a result of their increasing availability simplicity, security and improved performance. By 2005 however, many industry analysts including IDC and Ovum were forecasting a levelling of IP VPN demand and a corresponding rapid adoption of Ethernet services
Despite the tough economic climate over the past 2 years, both IP VPN and Ethernet services have been the bright spots in an otherwise depressed telecoms market. What are the reasons behind this growth? Will Ethernet eventually replace IP VPNs or will they continue to co-exist and which applications are best suited to one or the other? Our expert panel at NetEvents EMEA Press Summit will debate the pros and cons of both.
Introduced and chaired by Joshua Budd, Program Manager, Communications Group, IDC
Panellists: Craig Easley, Vice President of Marketing and Product Management, Accedian; Phil Tilley, VP Marketing EMEA; Alcatel-Lucent ; Kevin Vachon, Chief Operating Officer, MEF; Jeff Schmitz, Vice President of Networks & Applications, Spirent Communications
Alright, are we ready? A few brave people decided to stay; it's interesting to see.
You must be very passionate about Ethernet and IP VPNs or maybe the event organisers have bribed you a bit. Maybe there are free drinks after the event, I don't know. You're here and that's great. I hope to get a lively discussion going here. We have some great panellists who really have a lot of insight into the topic on many different layers and levels so I think we'll get a good discussion going.
But, before we dive into it, I'd just like to put our discussion in a little context and whenever I look at these technology issues, I think it's always helpful to stop and put yourself in the shoes of the customer, the end user, because we can debate the technologies and the details of the technologies. But let's stop and think about the customer themselves before we talk about everything behind it.
And I've identified in our research a number of challenges that enterprises are facing in managing their wide area networks and I'd like to talk about some of those challenges first, just in a generic sense, to try and understand how these challenges then impact the network and the demands on the network.
And if we're looking at enterprises, and this is not meant to be an exhaustive list of challenges but it's a few that I think would be relevant for the discussion today, we see a number of issues emerging in the market. One major issue seems to be hybrid solutions. There is not a one-size-fits-all technology out there, by any means, and the business needs are changing very rapidly. Consequently, enterprises are being forced to throw in all different kinds of technologies together to make everything work and this is adding a great deal of complexity to the wide area networks; a great deal of complexity. And I hope some of our panels can talk about the implications of that and their experience of dealing with that.
We're also looking at the workforce changing, so the people using these networks.
People are increasingly mobile and also nomadic and this is changing the way people interact and connect to the networks as well, and enterprises have to deal with this on an ongoing basis.
Security concerns, always a big issue. And from a network point of view and an enterprise point of view you want to understand, should we manage it ourselves or should we entrust it to a service provider?
Industry is changing. There are lots of services out there, lots of promises, but can the services always keep up? And what we're finding is that the technology promises a lot of functionality, a lot of abilities; but do the service providers, who are rolling out this service, get it to the market on time and are they keeping up with the changes?
Can they make a business case from it? And that's an issue as well.
And then, of course, just the way people are using the networks; the applications, the running over it. A network isn't just connecting sites anymore; people are doing so many things with it from voice to video, running security applications.
Managing this and making it work is not easy and, often, the managers of the networks will look at the technology to see; can this technology enable this and help me overcome these challenges. So, all these business challenges are having an impact on the network and the network is really, like I said, not just point-to-point activity anymore; it's becoming a world of its own.
We're connecting to the network in different ways from many locations. Networks have to connect over a metro area, national area and international. There has to be remote access to the networks and we're connecting different sites. We're connecting data centres to the headquarters, data centres to data centres so the topology of the networks is changing very much.
Applications were running over it. In some cases, VoIP is very popular but we're adding all kinds of applications on top of it. Video conferencing, security; all this is adding complexity.
How do we manage it? Compliance. This is a big issue particularly for certain industry segments. Financial companies are subject to a great deal of regulation. If regulations change, they have to implement changes within their network and they have to be able to do it quickly and apply it across the network; not easy to do.
And, then ultimately, the big issue is always cost. Networks are becoming more complex; this translates into higher costs in some cases and you get a situation: is the network supporting the business or the business supporting the network?
These are some of the issues that we're looking at in our research and I think would be relevant for the discussion today, and I hope we can delve into these in a little bit more detail. But, before we do, I'd just like to give you some indicators that we're tracking in the market. We do talk to WAN managers and we are seeing quite a few trends emerge and how these companies are using technologies and different kinds of technologies to address some of these challenges.
The topic of today is IP VPN and Ethernet; comparing them. There was a comment at the beginning: is Ethernet going to replace IP VPN? We can delve into that issue a little bit later. But the bottom line is IP VPN is widespread right now. The growth is levelling off. There's no question Ethernet is growing much faster; there's no question about that. But, right now, IP VPNs are widespread among companies and we're seeing that they're using IP VPNs for very specific purposes. A big one, remote access; IP VPN is very suitable for connecting all those mobile and nomadic workers that we talked about earlier.
And Voice; you can see over 50% of respondents in 2010 are running Voice over IP over their IP VPNs, up from 40% the year before.
Big growth in security applications, so you can see it's not just about connectivity; it's how they're using the network.
Let's take a look at Ethernet, and the picture's a little bit different. When I started looking into this presentation, I did come across comments on some websites, or websites saying: well, ultimately, technology has enabled the same kind of thing so if you're a company you should just be concerned about SLAs and cost. But our research is showing that's really not the case; that, yes, on one level they do enable the same kinds of things but companies are using them in different ways. So we see right off the bat the issue of connecting data centres to the headquarters. This is the big trend that we see among companies using Ethernet technology that we don't quite see with IP VPN.
Video conferencing is big again. High bandwidth applications are more suitable refining for Ethernet technologies; taking off a little bit faster than compared to IP VPN, let's say. But, ultimately, still the connectivity between the sites is the major purpose of this.
Well, what does it mean? And before I hand it over to the panel, I'd like to raise my opinion and we'll see how people react to this. But to get back to the point that was raised before - is Ethernet going to replace IP VPN? Well, I see these technologies co-existing. We already see companies having hybrid networks, combining technologies. Maybe not an ideal situation but it's what they're doing. I think maybe what we can consider is not is one better than the other but what kind of services are better suited on one network technology versus another? We can see connecting data centres, in my opinion, very suitable for Ethernet.
Managing your security internally. If you're a financial company and security is extremely important to you, you want to be able to manage it internally but with an IP VPN it's the kind of thing that maybe, if you're willing to trust a service provider, you can hand it off to them, and IP VPN is quite suitable for that.
Storage replication. Bandwidth-hungry applications in general we see more suitable for an Ethernet network, whereas Voice over IP could run quite well over IP VPN.
The picture I'm trying to paint is that we do - at least in my opinion and we'll see how the panel reacts - is co-existing technologies. Not one versus the other, one's going to wipe out the other, but technologies enabling different services depending on the type of the business and how the business is operating.
With that in mind, I'd like to start off with the outlook for Ethernet. I mentioned that Ethernet is growing very quickly. IP VPN is trying to level off but there's this big growth in Ethernet and that's definitely promising. Kevin, I'd like to ask you, before Ethernet can really even come close to replacing IP VPN if that is the case, there's an issue of coverage, connectivity. And we talked about some of the Ethernet exchanges; maybe you could tell us, what's the progress with Ethernet coverage? And if you're a company and you have international sites that you need to connect, how easy is that with Ethernet versus IP VPN?
Sure. Before I touch on that, though, I should say that the [method] is comprised of large service --.
Sorry, let me just interrupt. That's important, my mistake. We have Craig Easley from Accedian Networks, we have Phil Tilley from Alcatel-Lucent, Kevin Vachon from MEF and Jeff Schmitz from Spirent Communications. That's important, I must say. Sorry, Kevin, go ahead.
Our members are big/small service providers and equipment vendors. And people like the Alcatel-Lucents of the world sell solutions to meet both needs. A large percentage of our members are offering Layer 3 and Layer 2 services so, as an association, we completely buy into the fact that they're going to co-exist. We've never gone out there to try and suggest that this one is much, much better than this one; it's really just educate them on the positioning of both and the merits of Ethernet.
And that's serviced well, I think, in the industry [well].
I think we've talked about coverage quite a bit in the briefings over the last couple of days and the progress on the coverage front has been huge in the last year, driven by the availability of interconnect programmes, interconnect specifications that have been promoted heavily by MEF. We've educated dramatically and, of course, the rapid emergence of the Carrier Ethernet exchange market, which has made it much easier. So, I think if you look at the outlook for the market, with that coverage now in place that can definitely be an accelerator.
Another point I'll make about the coverage, Ethernet is growing faster than IP VPNs despite the fact that many incumbents actually got in the Ethernet business reluctantly.
They had a major investment in IP VPNs; they had trained sales forces and it was customer demand and competition which really forced them to. Some of them embraced it and moved forward with it to take as much market share as they could but others have been somewhat reluctant. With the coverage, with the customer demand awareness growing, you're likely to see that demand snowball.
You mentioned an interesting point; maybe I could ask you to just follow up on this.
You mentioned the competition in service providers went into it reluctantly; what seems to be the tipping point if an enterprise wants to adopt the Ethernet? Let's say they have IP VPN or maybe they're considering both; in your experience talking to the service providers, what sort of the tipping point that says: okay, I'm going to switch to this? Is it the size of the business? How they use it? Is there some trend that you can maybe identify? A sweet spot where, okay, we're going to take Ethernet?
Maybe, Phil, you're closer to the buying, and Craig; you see what people are buying and why they're buying it from an equipment perspective and exactly how they're using it, maybe it's probably a better question for you to answer.
Yes, it's a challenging question. It's challenging service providers. They're asking vendors: what is it if you're specifically looking about Ethernet deployment for delivery of services to an enterprise? Why would an enterprise buy one versus the other? To a certain extent, it's a desire to own the network yourself, to build the network yourself, versus to outsource. So, if the desire of an enterprise, for various reasons: political reasons, security reasons; the desire is to actually keep the network [build], the routing, the control of the network, in-house, specifics that are quite frequently a requirement for the local government, some financial organisations want to keep control of the IP routing tables, they will build a router inside an Ethernet network just buying Ethernet connections off the enterprise. However, if the desire is to actually go much more to say: right, I want to outsource that; I just want to look after and set governance rules and governance guidance and let the service provider run all the routing, and almost run and operate the cloud-based services, as it were, if we go that far, then they're more inclined to go to an IP-type network.
With Ethernet access, the one thing that we must be very clear is Ethernet is the preferred access option in most cases now.
I think it's simpler than that actually. I think enterprises are adopting Ethernet because we're damn clever marketeers; we've been inventing the next thing for the past couple of decades. Who, in here, remembers the eight-track tape? Then we went to the cassette tape because it's smaller so that was the differentiated --. We switched.
We got new kit. The songs that we bought on eight-track we now bought on cassette.
Then somebody came up with the clever idea that we need it on CD because you can hear this little hiss when you're listening in the [hum], so by marketing that and by pointing out all of the failures or the limitations of IP VPN, we're essentially moving the enterprise businesses to the next thing, which is Ethernet.
And you've heard all of the debates. Switches are cheaper than routers; switches are faster than routers; switches are less complex than a router; easier to manage than routers. Well, in reality, it's shades of better or worse and I think to Josh's point and the point that Kevin made earlier, there is a specific use for the two technologies in the network and certainly we have got rid of some of the inefficient users of IP VPN and replaced it with Ethernet services, but there are still things that you just absolutely can't do with Ethernet that require IP.
I must admit I'm not fully agreed because, actually, I think enterprise is one of my more virtualised cloud-based services. We've had the debate, the last two days, about the (inaudible) on enterprise to buy some sort of cloud application, or hosted applications. Actually, in a lot of cases, they don't really care how that's delivered but those that want to go and buy those hosted applications, it just so happens a service provider has come round to say: between large sites and data centres Ethernet is the best solution, and so it's the service provider providing a solution which is happening to be Ethernet in those cases.
If, however, the desire is to connect that cloud application to a home or residential location, a remote location, then IP VPN has much better scalability to those many thousands and millions of locations.
Did you want to --?
I'll jump in a little bit. I did a little bit of research and, [unfortunately], it was from Infonetics but (inaudible) they give us some data points in 2010, Ethernet Services grew at about a nearly 20% rate; IP VPN is still growing. So, despite a general depression in '09 and '010, both are continuing to grow but Ethernet about nearly twice the rate of IP VPN. And I think some of those are for exactly the reasons that you mentioned, Josh, higher bandwidth, etc.
I think another reason is, you brought out a point of: hey, if I'm going to cross carrier boundaries, can I get the kind of service and class of service that I expect? This has been a challenge but we, at Spirent, actually work with Alcatel-Lucent and on an eight-party carrier demonstration earlier in 2010 to show how the MEF is really helping standardise the services, the class of service and the quality of service and to start eradicating some of that, which I think will help Ethernet because, God knows, nobody wants less bandwidth so (inaudible).
I guess perhaps it's heresy to say, actually, I want Ethernet to be very simple to connect, become a transport, almost a utility, and that's okay for me provided I can then have the applications and support the applications on top. And we've talked about doing some of the application awareness intelligence stuff for me as an equipment vendor. Actually, Ethernet as a transport utility is fine because I think I can add enough value on top on the application awareness, but it's that interconnect utility interconnection that's --.
This is an important issue for me because is it just a utility or can a service provider differentiate their Ethernet service from another Ethernet service. And this leads to the question of the functionality of Ethernet, itself. Is it a mature enough technology that service providers can compete by saying: my service is different in some way than my competitor's service?
Absolutely they can, and that was one of the things that --. You're all familiar with the Metro Ethernet Forum [donor] that had the five value propositions of Carrier Ethernet and certainly a big hole that was missing at the time was the operations, administration, maintenance and the management, and so now that the OAM is there we have the tools in place inside the network to monitor the performance and [demonstrately] show that some networks are better than others. And using those key performance indicators of a high quality service, service providers are able to differentiate themselves and command value price for that model versus just continuing to slash costs.
And how do these functions compare to, say, similar functions in an IP VPN network?
You can do the performance monitoring at Layer 2 or Layer 3 and certainly the SNMP and RMON tools that have existed for IP, or continuing to be used to monitor those as well. But I think, as a transport technology, there was this perception that Ethernet's been fighting against that it's somehow less reliable, the value proposition of more (inaudible) were cheaper, came at some cost in terms of reliability or resilience as opposed to an SDH network. And with the kind of tools to do performance monitoring and demonstrately show that you can build a switch Ethernet network that's as reliable and high performance as the SDH network that's being replaced. I think service providers will continue to use technology to differentiate themselves.
One of the analysts that spoke about a year ago at a MEF meeting had done a survey on SLAs for Ethernet versus IP VPNs and the conclusion was they were similar but what they also found was that the product teams in some of those service providers actually were reluctant. They felt that the infrastructure was actually better for Ethernet, or the infrastructure provided a capability for a much stronger SLA with Ethernet than what they were actually offering, and there was resistance in some of those companies to actually creating an imbalance of the IP VPN teams, etc. The Layer 3 teams were saying: we can't offer a better SLA for Ethernet; all our customers are going to want to switch, and so and so forth. But the capabilities are there and it's really now about (inaudible) it's about the management side of it now, just coming through all the fancy portals and management statistics that are essentially available now or becoming available.
And some this comes down to purely a generation of technology. If we look to some of the IP VPN services that are on offer today, they're getting on for 10 years old.
They're built on an old generation of platform, whereas obviously now with the Ethernet services a lot more are being built on newer platforms, and especially those who are offering converged Ethernet and IP off a new platform with QoS capabilities, management capabilities end to end. And certainly that's where we, at Alcatel-Lucent, say we've come in as a third generation of routing platform, which is service aware, which is combined Ethernet and IP. And therefore what you're seeing now is the Ethernet networks having more capability than what the old legacy IP networks did have.
Just to make another point. I was speaking at a conference last week with one of our member companies and there was a smaller service provider (inaudible) Ethernet player, and they had a bunch of application service providers at this conference who bundle the application with bandwidth, and what they were saying was, their customers that they're selling the applications to actually really like Ethernet and preferred it.
Coming back to the coverage issue, in the past, when they've tried to - especially out of country, it was a smaller country - when they've had to go out and buy an international Ethernet service, they've found that it was still quite expensive. And the view is that, because they didn't have the connectivity, they weren't connected to an exchange, they had maybe some limitations as to what partners they were going to use in a particular country. So the view is that with the coverage and the accessibility to different partners' access services growing, then those price points should come down with competition so that will, in turn, further drive that business.
I might add the quote from an old movie. For those of you who saw Jurassic Park, you remember there is no reproduction in the park because they were all women and, ultimately, some of them converted, and I think Ethernet is like Jurassic Park is. In that movie, it was life finds a way but I think what we've learned over the last 10 years is that Ethernet will find a way and, ultimately, that's why I think Ethernet's growing at double the speed of IP VPN. We're going to have hybrids for a long time but Ethernet is solving every challenge that's raised to it and, long term, it's going to win; it just didn't get killed right away like a lot of people predicted it would. But I think, over time, you're going to see Ethernet. But we see a hybrid world probably for an extended period of time and from a test (inaudible) complexity is always good because it creates uncertainty and certainly creates the need for testing so that's not all bad. But I think that's just the way it's working.
You're testing a lot of these networks; is the performance living up to the promise? I think I agree with these guys, Ethernet's going to find a way. I think there are challenges in Ethernet right now. I think that management is getting better. I think the ability to interconnect cross country will get better. I think that MEF has done a good job of trying to move that forward. We've been part of that with Alcatel in sponsoring some and EANTC in sponsoring public [testing] to show that it's possible, to move the bar forward in the industry, so I think we'll get there. I think eventually we'll see this trend continue. Ethernet will probably double -- it's already bigger than IP VPN from a dollar perspective and I think that will continue.
(Inaudible) the backhaul. Craig, you're very active in Ethernet and the backhaul, how is performance in the backhaul? Do you see any major challenges? In the Status there's a massive rollout of fibre to the base stations; we don't quite see that as much in Europe just yet. You disagree, yes. How is that coming along in The States?
As part of the briefings that I've been doing over the past couple of days, I've been sharing the story of a Tier 1 mobile operator in the States and their rush to rollout the 3G and 4G radios required them to also rush to deploy Ethernet out to the base stations, and rather than build it themselves they partnered with wholesale access providers. And what they found by monitoring all of those links was there was a huge diversity in the performance that you get from one wholesale access provider and another, so without those tools to measure and keep those providers living up to the SLAs that they had offered, the rollout would not have been a success. And there are three variables there: you have the equipment that it's based on, the media that it's carried over and the engineering expertise that architected and designed the network and they're very different. For some people it's all fibre and all on new equipment; some is on bonded copper and using legacy media and non-optimised Ethernet
equipment. Some of it's built by engineers that have PhDs and staff of 1000s and others are built by some guy that picked up a PDF off the Internet that [said] Carrier Ethernet.
The one thing I would say in mobile backhaul, especially in North America, actually it's all built upon Alcatel-Lucent and if you look at the results that came out today, we grew our IP business 63% year on year, if you actually read through the results you'll indicate that we're actually sent -- a lot of that growth is driven by mobile backhaul, especially the Tier 1 operators in North America; we've done massive, massive rollout the last six months of that. And, now, where are we seeing that? Yes, absolutely, we're starting to see that happen across Europe. It's a slightly different pace because of the LTE process and also it's a mixture of Ethernet over fibre and Ethernet over microwave.
That's an interesting point right there. How does that affect performance?
Again, there are different engineering challenges with microwave clearly atmospheric driven but the good thing, and certainly the MEF (inaudible) work with -- we've defined MEF services that will go over any access technology, whether that be fibre or microwave. It has the same attributes, the same look and feel. For sure, one has to think about engineering the network slightly differently because you've got new atmospheric conditions to worry about but I think with the new adaptive packet radio [stub] those challenges are being overcome and we can build an effective, robust, reliable aggregation Ethernet aggregation network using microwave.
Interestingly enough, fulfilling the backhaul need by Ethernet, I think anyway, is going to drive business services growth. I'll give you a couple of examples. Talking about the US, there's a massive push to quit fibre. The Verizons of the world are putting RRPs out and hundreds and hundreds of small telephone companies in the US - there are 1000s of them, rural telcos - are trying to bid on these things sometimes as state network consortium, most of which just a bunch of stimulus funding, so they're now trying to figure out how do we become compliant to grab this piece of business that's happening, so they're looking at (inaudible) standards and building out Ethernet offerings. Where they were selling point-to-point services mainly, now they're saying: hey, I've just had the bill that an Ethernet offering for backhaul, now we're going to start Ethernet business services. They actually don't want to be in the business of selling Layer 3 services because the smaller guys don't necessarily have the people to
offer that kind of managed service; they're happy to offer the Layer 2 service.
Another example. I've talked to quite a number of mobile operators whose names you wouldn't think are synonymous with being a mobile operator and they're saying: we just put this huge country infrastructure for transporting all this new data and we're going to sell business services too; it's the same infrastructure. Then mobile backhaul is going to drive Ethernet Layer 2 growth as well so it's a bit of an odd thing.
Okay. Well, it seems that the coverage and the infrastructure issues are being addressed, but again you still hear comments in the press (inaudible) performance.
What do you guys feel about running applications over IP VPN versus Ethernet? For example Voice; this is, I would say, a common VPN service. Are there issues with jitter? Do you see differences with Voice over IP VPN versus (multiple speakers)?
Yes, if I'm perfectly honest, the way we're building networks, it's exactly same, whether it's Ethernet or IP. Actually, the engine that's processing the packets is the same engine with a slightly different interface on the front. Performance-wise, we can engineer the network quite happily one way or another. The difference is in the signalling, not in the packet processing.
Alright, we have a few minutes left. Do you guys still want to make --? We forgot your opening comments so I still want to get to them. They have some comments that they'd like to make; I've mixed this up quite well. But before we get to some closing comments, we can take some questions from the audience if there are any. I don't see anyone rushing. Are you not as passionate about IP VPN and Ethernet networks as I thought? Okay, in that case, would you guys like to make your closing comments, if you have some closing comments to make?
I think we all probably took advantage of saying what we were going to say anyway.
But maybe, in closing, I think you had a good summary of why people choose maybe one over another. I think there are still challenges for Ethernet but, clearly, we're making progress every day and hence you see the growth of Ethernet. They're going to be in a hybrid roll for a long time and Spirent remains dedicated to help both the IP VPN and Ethernet Services side in terms of testing both for the network equipment manufacturers and the service providers.
Great. I think the great thing is everybody simply puts, and I think it was even done on the top of this debate, Layer 2 VPNs versus Layer 3 VPNs; and it's the 'versus'.
And the one challenge we face, as an industry, is to cross out that V and say 'and' because they both sit together; it is a hybrid, especially if we look to cloud and where we're going in cloud and we've had lots of discussions, as I said, on the cloud. There is no cloud without network. The network can be, and will be, IP predominantly to the residential over next generation access DSL, even mobile, and that will be IP to a gateway and then access into an Ethernet data centre. Between data centres it's Ethernet; they're both robust technologies, they're both sitting there.
And I confess that, for the 20 years of my career, I have been pushing the adage switch where you can, or out [where you must] but it's a lot easier to drive the nail into a board with a hammer than it is with a screwdriver, and I'll leave you with that, too. Each of those tools has their job and they do what they're good at. I think the explosive growth in Ethernet is rooting out a lot of places where IP and VPN were used inefficiently in the network just to make those connections that didn't exist. The (inaudible) work on the wholesale access services, definitions and the wholesale access certification programme that's going to be coming down the pike later this year will also aid those international Ethernet cross connects.
I think I made my opening comments earlier that MEF takes a position. If they're going to co-exist we don't really promote one versus the other.
Sure we do.
Vendors do. We've actually got a white paper coming in the next couple of months, Layer 2 versus Layer 3. I shouldn't have said Layer 2 versus Layer 3. But I do want to make the point that Phil made about the cloud; I think that's going to be interesting to see how it all plays out when the dust settles there, to see if there's a trend one way or the other. We think it's going to be Ethernet; I think most people think it's going to be Ethernet; it'll be fun to be talking about this in a year's time.
Okay. Well, on that note, I think we actually have a fair amount of agreement; not so much of a debate but I think there were some interesting issues raised. It's getting late and thank you very much for your time. Thank you.