Executive Exchange, Frankfurt 2011





Ethernet Exchange: Open for Business!


Mark Lum, Co-Founder and Market Lead, Layer123

This is the clear message from the 2nd Executive Exchange hosted recently in Frankfurt by Layer123 on 12th May.  The first, 6 months ago in London, saw the market players rubbing shoulders and jostling for position as they set out their Ethernet Exchange stalls.  Now, it is time to find out how the marketplace has been developing and who is buying...


From left to right on the Executive Exchange...

Erin Dunne, Director of Research Services, Vertical Systems Group

Mark Cooper, Product & Marketing Manager, Interxion

Adam Janota, Director of Global Networks, Equinix

Lutz Blank, Vice President, Marketing, Colt Technology Services

Chris Eldredge, President & GM Ethernet Exchange, Telx

Nan Chen, President, CENX



The big questions… “What is an Ethernet Exchange?”...



What are the future prospects for market size and growth? Who is going to use these services, who are the early adopters, and who will be the larger users, later on?  How will the ecosystem make money from this new type of Exchange?

The air has cleared since the 1st Executive Exchange and the leading players seem much more aligned on what an Ethernet Exchange actually is: let’s hear from them in their own words…


“It is really interconnect points for the carriers where the Carrier Ethernet service is being exchanged.




And you want to do it with scaleability, reliability, and quality of service ensured… that’s the definition from the (MEF) global interconnect group.”


“Telx views Exchange as a tool that is necessary to bring all the customers that are in our data co-location facilities together.


So if you look at our whole ecosystem, we have about 880 customers that sit in our data centres. We utilise traditional interconnection products whether it is SDH, SONET, E1s, T1s.  People interconnect in our facilities and we utilise the Exchange just to bring all these folks together and provide true value to simplify the process.”


“As a carrier, we are the odd ones out at this table.. Colt’s view is that what we are doing to some extent could be like a distributed Exchange across Europe.  In terms of what we are doing for our customers, we have a big Ethernet network, the customers come to us, for example connect to us in Frankfurt, and we provide them with A-B connectivity to any location in Europe, either through our fibre network where the B end is also on our fibre, or through the help of third parties.


The difference is, we keep end-to-end control of the commercial package that we are offering.. so from that perspective, we are a very un-neutral, biased Exchange!”


“Equinix is a co-location data centre provider which is also an interconnection, and from our perspective, Exchange is an evolution of interconnection.


Similar to what happened to internet exchanges a few years back, it’s the ability for customers to connect their layer 2 services in a transparent manner.  This is the first stage of that evolution and as Ethernet technology is growing and is replacing other technologies for ubiquitous interconnection around the world, if we can help that model to become more transparent, that’s what we want to do.  Being in the centre of carriers and other companies taking advantage of the interconnection, we felt like it was our duty to solve that problem.”


“We have a similar but slightly different view from the others in that we don’t operate our own Exchange: Interxion works in partnership.  We see an Ethernet Exchange as a complementary method to the ways we normally do connectivity within the data centre.


The most interesting thing is - that no-one has mentioned so far - it’s an ideal platform to build a community and a market place around.  I think that’s where there is a real opportunity for the Ethernet Exchange business – it’s not just about hooking carriers up together, it’s the business you can generate around the Exchange and the kind of ecosystem you can build on top of it.”


“Well, I am just shocked!  Because Vertical Systems was part of that debate [the Executive Exchange] last year and there was almost a fist fight over the word "carrier neutral".  Because it was up there in the definition and with the just-announced acquisition of Tinet by Neutral Tandem, it got pretty heated!


No one is using that buzz word "carrier neutral" any more.  When I put up my definition we say "open", because I was not sold initially on the whole carrier neutral thing, because if you are an Exchange provider and you offer the ability to get your customers from point A to point B, and it's there for the choosing whether you are offering it, or your partners are offering it, that's an open solution and doesn’t need to be neutral.


So that was the big debate last year, so its kind of interesting that we have had some movement on that.


[ed. note: various Exchange providers interject that they are still neutral]



Ethernet Exchange has burst onto the scene and the market is still catching up: some FAQs, selected for space…



“What actually is exchanged at the Ethernet Exchange?”



Services, data traffic, information, information services, it is like a traditional point to point service. What goes over that?  IP...it depends upon the application that the provider or the buyer is using.  It varies..


It's a Layer 2 interconnect point, so to the Exchange provider it’s two pipes and if you throw gummy bears on them, it’s not necessarily our concern, and a variety of clients will send different kinds of traffic. One client may send either voice over IP traffic and another client may send their video traffic.  For us, it’s all about the Ethernet layer.



“What about traffic profiling?”



What gets exchanged is a fixed amount of bandwidth.  On any individual transaction the customer orders whatever is on offer, and the end customer and the intermediate operators will all do the commercial transaction on that fixed nailed-up bandwidth and there is no statistical multiplexing.  That bandwidth is then available for the end customer.



“So when CIR, PIR, statistical multiplexing come into play.. what gets exchanged?”



At the Exchange point we don’t necessarily multiplex, or oversubscribe.  In terms of what a customer will do, it is their business and from the provider perspective, it is also their call as to how they want to manage and configure the pipes, for example a 20 Meg EVC between the two providers.


We see this quite a bit... many people, because they are familiar with a TDM-based service, say "I want a real time service" (and a typical example are the wireless backhaul carriers) .."I know T1 and E1, I want to have the same thing, just cheaper, quicker, faster over Ethernet".   But they are wasting a lot of money in that particular regard: they could have done a split between real time and best effort traffic.  It could save them 50%, but they say well, I want it real time.  So I think statistical multiplexing is still not well understood by a lot of people, and more education is needed.



“What is the model for the Ethernet Exchange?”



CENX: There are multiple ways of doing it: you need to match the service profiles. For example, if two customers have two different burst sizes and come together, if you want to get packets through the Exchange with a smaller burst size on the receiving end, what are you going to do?  You can drop it, or you can shape it.  So basically you ask the carrier, which one would you like to do? These are the functions the Exchange could do to help interconnect between the two carriers.

Equinix: It is taking whatever comes in at one end and putting it out the other end: it's an evolution of interconnection as I said before. It's in customers' best interest to agree on the policing of the traffic, or the dropping of traffic packets and I don’t think as an Exchange you want to be in the centre of it since its switch capacity far exceeds the traffic on the ports - so there is no contention.  There is no reason to control it, because the commercial agreement still sits between the two buying and selling parties: "I am going to buy a 100 Meg over a Gig service, or 100 Meg over an oversubscribed 100 Meg port.. in that case I am sending more traffic, and I am willing to take that risk".  I think that if you add another third party complexity in the centre, you are asking for trouble. Because now there is going to be a point where someone is going to say "well it wasn’t my network, it was the Exchange, that had a policy that dropped the traffic".

CENX: We know the technical details. That is exactly what we do at the beginning of each interconnect point. We work with them... we actually study each and every single service provider, service profiles and service types, for every single one of them.  So when you actually do the mapping between the two, you know exactly what needs to be done, and it can be agreed upon between the two. That’s what we do today.



The Business of Ethernet Exchange: Final Words


I would ask the basic question, which is, "We have been in this market for almost 2 years now, where is the revenue?"  It is a very simple question, right?


And I think it's a bit premature, because it takes a while to get this set up and running.  You need to get the facilities up and running with space and power (some folks here are already ahead of the game), you have to get the correct mix of buyers and sellers out there who are willing to do business in this kind of niche market and then you have got to get the traffic flowing.  And that’s not a simple task with all the trappings that you have around that, building portals and the business/technical agreements ..this takes a while.


So in a year’s time, when I say "Where is the revenue?", there will be some expectation that it is coming!


I think a lot of people are just saying "wait and see". Particularly as there is a number of different Exchange platforms - are people willing to deploy with Telx, with Neutral Tandem, with Equinix, with CENX?  I think there is still an element of people waiting to see who's going to be the big gun before they commit.


I think the other interesting thing is the way the Exchange has gone in different directions. The Neutral Tandem model is quite different now, you now call yourselves EtherCloud, you can access the platform from anywhere; you have Equinix who have already moved forward from just focusing on carriers, they are looking at Cloud, they are looking at Enterprise, they are looking at mobile; and Telx who are focusing on what was their core business, what they were really good at; and then you have CENX who focus more around simplifying the interconnection process and building the online market place.


So it will be interesting to see where and why they have gone in four different directions even though they kind of started with the same idea.


So it’s a new product for the industry and for everyone, there are a lot of options when you start it, and adoption does take time.


As you on-board more clients (we are all on-boarding clients as we speak), it does take time and now it takes a few months to set up, so these initial NNIs are very similar to point to point NNIs, we have to on-board the client, we have to verify their service, we have to interconnect with them, test with them and so on, then we have to train their teams to use them, and as we said there is a challenge with buyers..


The sales people are very much incentivised, so they will look everywhere and anywhere for the opportunity and they are much more aggressive, they are much easier to convince to do something new.  I think the procurement teams who actually buy are people that don’t have as many incentives, are not commission driven (often), they have a very set amount of procedures that they use.  So for them to learn a new set of procedures, without immediate benefit, it is sometimes difficult.  So there is a revolutionary and educational process that takes place.


The reason that we are one of the early adopters, one of the early connectors to these Exchanges is that we fundamentally have an interest on both sides, the buying and the selling side.


So lets look on the buying side..  We are a regional and European operator, so for us the story line makes perfect sense. Let's get connected with some Ethernet exchanges and that will make life a lot easier for us to get connectivity to the rest of the world.  Yes it is early days..  When we have small buying volumes at the Exchanges it could make life easier for us.  On the sales side it is the same thing, where we have our big customers they are doing the business with us (and we are doing it with them).  So I think for moving forward, the obvious next steps are for small volumes, but the small volumes aren’t going to make this market fly, however would be nice to get going!


The big question for all of us, is not to focus on the technical side, but to really focus on the next few months on the business models and business discussions.  How do we make this work at a commercial and operational level, so that both the buyers and the sellers with the Ethernet Exchanges in the middle see the real benefit.  That part - the business side of it - seems to be still a bit slow at the moment.


We started in October 2010 and the revenue has definitely behind what we anticipated, but if you look at the amount of carriers and providers we have signed up over the last 6 months, it is continuing to grow each month.


We know we are not going to replace direct NNI (we know that there is a market for that) but there is definitely also a market for Ethernet Exchange.  But we are starting to see more buyers come on to the Exchange, initially it was all sellers quite frankly, but now we are starting to get traction with buyers and that actually is translating into real revenue and each month we are starting to grow and we are creating the market.  It is not only bringing someone onto the Exchange and hoping they will use it, it is about bring the buyer and the seller together.  We have the sales teams focused on bringing these folks together, creating value, meeting with customers and clients.


So I think we are starting to get more traction and I think next year when we sit around this table, the revenue numbers are going to be a lot better than we have anticipated.  It's there, it's starting to work and it wasn’t as fast as we would like it, but with any new product, it typically takes time.  In my prior life as a product guy, rolling out dedicated internet access, it took 18 months to get traction on it and that’s internet access! - so I mean this is definitely a new product and we are starting to see real results.


I think the key focus for all of us is to figure out what is the best way to get the business going, to get business and on the track of growing every month.


And also from a carrier’s perspective, we want to make sure carriers work out the business case.  You have got to find the place where you think the Exchange will generate a lot of volume.  We feel we have found that place and we are working hard on it, but with a new product it takes a long time. I remember starting back with the MEF's Carrier Ethernet 10 years ago.. now look at today: the momentum in unbelievable!


So this is going to take time, but in general I always feel that the Exchange is a way to scale the industry in general, not just for CENX.



(any errors and/or omissions are the author’s, please also refer to speakers’ original presentations)