Roundtable - The Road Ahead

15 June 2010 |


Capacity invited a number of senior figures from companies with a stake in the market for carrier Ethernet.

PARTICIPANTS:


             
                                                                                                      



In a few short years, carrier Ethernet has infiltrated from its origins as a LAN standard into the metro and long-haul networks of carriers and service providers all over the world. Most market watchers believe that, ultimately, carrier Ethernet will fully supplant legacy alternatives thanks to its qualities of easy scalability and cost-effective provisioning. But for the meantime Ethernet must work alongside existing infrastructure. With so many carriers at a halfway stage between a pure Ethernet world and a legacy past, we ask our panel about the challenges entailed by a hybrid legacy/Ethernet strategy. How fast should carriers look to migrate to all-Ethernet? How do you avoid newer Ethernet-based services cannibalising existing revenue streams?

MULDERS: It is clear that customers are embracing Ethernet as a cost-efficient methodology for more scaleable, higher speed connectivity. In the short term carriers who are moving to Ethernet need to operate another infrastructure. An issue might arise when they make different technology choices than they would have if they did not have a legacy infrastructure. An obvious challenge is the cannibalisation of existing revenue streams. Ignoring it does not make sense. Retention programmes will most likely involve offering more bandwidth for the same, or slightly reduced, price. Existing networks, like SDH/Sonet, can still be leveraged, as EoSDH/EoSonet, to generate new revenues. In time providers will have to optimise the cost structure by migrating to a fully native more cost-efficient Ethernet service.

SCHWARTZ: Take-up of our Ethernet offer locally, nationally and globally is growing very well. We’re experiencing robust top-line growth in all areas, fuelling incremental sales across our product portfolio. Some of the growth is coming from customers converting from legacy to non-legacy, and some is augmenting legacy infrastructure with next generation. It’s not necessary to move everything away from legacy infrastructure on the spot. We find it’s useful to be able to meet customer requirements for a hybrid approach – for those not looking for a clean swap out. It’s about managing the transition and making it easy for customers. We seamlessly manage that. That’s the value of being a global Tier 1 – we can supply to meet all types of service, from cloud to storage area network capability. Our size enables that. With a pure NGN approach you can’t take a soup to nuts support role.

APPLEBY: It seems that maintaining a hybrid infrastructure would present some cost and support challenges. However, our discussions with carriers indicate that they either have transitioned to Ethernet or they are aggressively pursuing migration to Ethernet. Given the inherent cost and scalability benefits associated with Ethernet, it is logical that network service providers would pursue migration to Ethernet in a measured way to achieve the best balance of cost reduction or revenue retention.

BROOME: Colt is focussed on business services and is not encumbered with the challenges of transforming a legacy consumer network. This means we have been able to be very dynamic in terms of rolling out new infrastructure optimised for support of Ethernet services and transport for IP services.

Traditional leased lines are still an important part of Colt’s business, and Ethernet services are an opportunity for our customers to choose rather than be forced to migrate. As more than 80% of Colt’s top 500 customers already use Ethernet services, we are well on the way to an Ethernet-powered transport network.

 The biggest drawback that carrier Ethernet has faced to date is the difficulty of getting two different Ethernet networks to interoperate without a great deal of preliminary technical and commercial activity. The expense and time that it takes to set up an interconnect arrangement between networks has been prohibitive, especially for smaller less well capitalised service providers. Now the Metro Ethernet Forum has moved to solve the problem with standards that will create a common interconnect language. So is the much anticipated eNNI standard really going to have a major impact on Ethernet popularity? What is MEF’s ideal future role going forwards?

SCHWARTZ: Verizon is in a unique situation here. We play a leading role in MEF, at board level. We’ve been deeply involved in the whole standards process. But added to that we have our own stringent certification process for interconnecting with other networks all over the world. No single carrier has infrastructure of their own in every country in the world which is why in spite of our large network we still need relationships with 130 or so off-net partners. They all have to go through a robust process to meet our standards.

MULDERS: Yes, it will have a major impact – especially for international network operators. When eNNIs are standardised, the roll-out of a uniform Ethernet service will go much faster. It is like an E1 – it’s the same everywhere. The time that parties need to technically interconnect is significantly reduced. However, the commercial and legal negotiations still need to be done, which is significant if a carrier anticipates an international Ethernet roll-out. Using an Ethernet exchange network, like our GEX, will significantly improve the time to market and allow parties to make use of smaller, more competitive carrier Ethernet networks.

Expereo is very enthusiastic about the efforts of the MEF. But while we support the efforts of the MEF, we also understand that the efforts of a standardisation body must also be balanced against the needs of users who drive and create advancements within any technology.

BROOME: MEF’s specification for eNNI has been recently ratified, but Colt has been anticipating these specifications in the many eNNIs which we have already deployed. The role of the MEF is important as an independent body defining Ethernet services and allowing the interoperability of end-to-end services across operator boundaries.

Of course it’s up to operators to guarantee performance, and Colt does this by deploying demarcation devices at the customer’s site for off-net connections giving end-to-end service visibility. While Colt has over 100 partner operators, we also offer Ethernet services to our wholesale customer base. This allows other operators to take advantage of Colt’s Ethernet reach using our Ethernet Hub & Spoke product, which is our wholesale eNNI offering.

APPLEBY: eNNI alone will not have a major impact on Ethernet popularity as it is already widely popular in the LAN space and quickly gaining the same acceptance in WAN and global networks. However, a standardised eNNI will definitely create an impact and will help achieve the MEF’s mission of expanding standardised Ethernet services across the globe. MEF’s ideal future role should remain unchanged from its current role. The member forum community has and will continue to be an integral part of defining Ethernet standards, liaising with other standards bodies and educating the marketplace about the benefits of Ethernet services.

In what looks like being a truly seismic year for the development of carrier Ethernet, we are now seeing carrier-neutral exchange facilities spring up that allow for traffic to be exchanged between Ethernet network operators via one simple contractual agreement with the exchange operator. The model has already been proven to work well for internet traffic, but will it apply so well to Ethernet? How revolutionary are the carrier Ethernet exchanges currently appearing? Will they change the Ethernet market? Who do they best serve?

APPLEBY: Carrier Ethernet exchanges aren’t revolutionary but are the result of evolutionary technological and service developments which will complement the fast-growing Ethernet market. They may contribute to market acceleration in certain areas and speed up time to market for providers looking to enter or simply achieve service and network extensions. Ethernet exchanges primarily serve Ethernet service providers looking to eNNIs as a vehicle to either buy or sell their Ethernet services. Secondarily, exchanges may appeal to certain enterprise clients who can easily access a multitude of Ethernet service providers, whether to purchase transport or partner to deliver their own application services.

SCHWARTZ: These exchanges are going to allow smaller players to get easy access to local facilities in regions where they will never have a network. Now a service provider on the US west coast can easily get to the mid west or south coast without having to work through difficult negotiations. For us it will help fulfil some very specific needs, but we will generally be making use of arrangements that are already in place.

BROOME: Carrier Ethernet exchanges are certainly a new way of interfacing between operators, and present an alternative way of interconnecting for their existing ranges of Ethernet services. As a pioneer of Ethernet interworking, Colt welcomes initiatives that help make Ethernet more accessible. Carrier Ethernet exchanges provide a useful way of adding new coverage and routes to market but it’s important to note that it’s the carriers themselves who are in control of the end-to-end service and can offer the end customer a true managed Ethernet service. This is because carriers are responsible for the Ethernet service SLAs.

MULDERS: Yes they are revolutionary, because they will create a uniform global network, allowing parties to easily obtain connectivity, to any site anywhere on the planet.

Network transmission speeds used to be measured in Mb. Only a few years ago, the ability to send data at 10Mb per second was right at the bleeding edge of network capability. Now we’re seeing network operators talking about being “ready for a 100Gb world”. It can’t be long before we’re talking terabytes. But before that happens, the industry has to decide where it jumps to after 10Gb. The usual rule is multiples of 10. But 100Gb is not seen universally as a commercially viable performance level, many seeming to prefer 40Gb as a compromise. Are those that take a more progressive view right? Is 40Gb Ethernet a false move? Should we all be looking to jump straight to a 100Gb world?

BROOME: Demand for higher capacity connections will be driven firstly by operator backbone requirements. Higher speed connections will make their debut in the core before volume uptake by end users. One of the hurdles to overcome is the price dislocation for user interfaces between the 10Gb ports available on Colt services today to the 40Gb and 100Gb ports of tomorrow. 100Gb is the ultimate aim for Ethernet speed, however we are yet to see how the economics will pan out and both may co-exist in the medium term. It’s important to make sure the core network is ready for higher speeds – something which we’ve done by investing in a new DWDM backbone network delivered across our network in June 2009. This network is used to deliver high-speed customer connectivity up to 10Gb Ethernet and is 100Gb ready for future upgrade.

MULDERS: We’ve seen bandwidth needs exploding – both in access and backhaul. Key drivers have included video. This will not stop – in fact it has just begun. It will not take long before 100Gb is dead. So the answer is yes. And if everybody is jumping on it, the technology becomes relatively cheaper even. Like Gig lasers for FTTH are cheaper than 100Mb lasers.

APPLEBY: This question is probably best answered by the equipment manufacturers and service providers. In my opinion, both manufacturers and service providers will have to weigh the benefits of deploying a solution that may quickly become obsolete. Equinix will work to accommodate our customer’s infrastructure needs in our International Business Exchange and provide relevant interface options with our interconnection services. 

SCHWARTZ: Customers that once wanted 10Mb now want 100Mb, and customers that wanted 100Mb now want 1Gb. We’re starting to get lots of 10Gb requests coming in, so that’s on the way to becoming mainstream. We’re not getting too many serious queries yet about 40Gb or more, but discussions about that are starting to happen. Where - demand is going, we will be leading the way, that much I can say. We will continue to push to provide the bandwidth that’s wanted. Verizon was the first carrier to successfully deploy a commercial 100Gb ultra long-haul optical system for live traffic, deployed on the company’s European optical core network between Paris and Frankfurt.