Ethernet services

Ethernet Expo: Gurus Grade Themselves

NEW YORK – Ethernet Expo – While carrier Ethernet is taking off as an idea, the service is still undergoing some growing pains, at least in the United States.

Some of the stumbling blocks were noted in a session here titled "Enterprise Services: A Moving Target." Panelists noted some of the wonderful things anticipated about Ethernet services, and went on to explain what's missing from today's offerings.

The most detailed outlook came from Richard Klapman, (NYSE: T) product director for converged packet access. Klapman issued his carrier Ethernet report card, speaking not as a provider, but as a buyer. AT&T offers carrier Ethernet over its own fiber but also resells other providers' carrier Ethernet in some cases, and a request for proposal (RFP) earlier this year yielded some disappointing results.

Klapman gave carriers a B+ when it comes to availability and pricing. "Ethernet's really not available in all their LATAs or all their buildings," he said. "This gets in the way of carrier Ethernet becoming a multibillion-dollar market."

And prices remain higher than what AT&T's customers are expecting for Ethernet, he said. Klapman also took issue with the pricing process, which he found too slow, grating on the supposed ease of working with Ethernet. "It doesn't go with the simplicity -- faster, better, and cheaper."

For interoperability and performance, carriers get a C+ grade from Klapman. Speeds and interfaces often don't match, AT&T has found, and many carriers don't guarantee performance at all, making it difficult for AT&T to pledge QOS for its own users. "It's kind of like Ethernet's been out there for three decades and everybody's implemented it in different ways."

Klapman noted that interoperability should be aided by the Metro Ethernet Forum (MEF)'s Carrier Ethernet Certification Program, which will set the bar for Ethernet services. (See MEF Adds Carrier Certification.)

One caveat: All this applies to AT&T's U.S. shopping experience. "Ethernet is actually further advanced in the Pacific Rim and Europe and Australia," Klapman said.

Still, other panelists noted that Ethernet could use some more functionality to serve enterprises -- items such as monitoring, for example. Panelists discussed the possibilities for Ethernet to serve in storage markets, but storage networks often come with a low tolerance for downtime and a need to retrieve information immediately. "[Customers] need visibility at every step in the network," said Stephen Barr, director of carrier partner business development at Ciena Corp. (Nasdaq: CIEN). "We need to turn Ethernet into more of a deterministic service, more of a guaranteed service."

Audience members got a taste of an enterprise wish list from Chris Lin, vice president of infrastructure for Ellie Mae Inc., which runs an online service connecting mortgage seekers to prospective lenders. Lin explained how the mortgage lending industry has taken to using networks to corral the mass of people and documents required for each transaction, then noted some features he would be willing to pay for.

They're the kinds of features that are available, but not at the level Lin wants, and not necessarily from the carriers providing Ethernet services. One was document management -- the scanning and storing of the voluminous paper documents that still permeate the mortgage process. (Lin said he's heard of one company that outsourced the scanning of documents to India.)

Another was data replication, which Lin said needs to be faster. "I cannot have good instantaneous replication, and yet because of disaster recovery, I have to keep things as far apart as possible," he said. "That's driving me crazy."

Neither case necessarily relates to the Ethernet argument, but Lin held out this carrot: If such features were available, that might "push me over the threshold" to buy other carrier services, he said. "The advantage of outsourcing is that I can gain additional services."

— Craig Matsumoto, Senior Editor, Light Reading

digits 12/5/2012 | 2:57:37 AM
re: Ethernet Expo: Gurus Grade Themselves So Ethernet services aren't yet cutting it in the US. How much of this could be down to the carriers needing to squeeze as much out of legacy services for as long as they can, while still being seen to be doing *something* to meet the demands of Ethernet hungry customers?
desiEngineer 12/5/2012 | 2:57:36 AM
re: Ethernet Expo: Gurus Grade Themselves I think providers do themselves a disservice when they model Ethernet as a cheap service (because it's ethernet and ethernet's cheap). It's kind of like Dell getting ahead of themselves with free shipping and other bottom-line-depleting maneuvers that were completely unnecessary.

E.g., ethernet also provides a convenient network model, and subscribers should pay for that convenience, shouldn't they? Or are they going to sit there with their Frame circuit forever?

Ethernet is a service, and services cost money, and new services cost a bit more.

Undercutting to get business puts a tremendous strain on the viability of the service. It becomes a self-fulfilling prophecy of doom: ethernet services die because they cannot beat the price model they are assumed to fit into, but which they don't.

Recently, this seems to be typical of all provider offerings: undercut, get the business, stall because the revenue isn't enough, blame the service, move on to next year's technology in an attempt to stay alive.

Q: Is this typical of every industry? Every silicon based industry?

eramk 12/5/2012 | 2:57:34 AM
re: Ethernet Expo: Gurus Grade Themselves I don't think its fair to blame the SPs. It was the vendors who introduced the market to the notion of Ethernet's "new service economies", while choosing to ignore the scaling complexities of L2 VPNs, and the massive legacy network integration costs needed to extend Ethernet services' reach.

The era of Ethernet Services is largely coming to an end and the MEF Service certifications are likely to be the final nail in the coffin (the last thing we need is proactive commoditization of services). From a SP perspective Ethernet will be relegated to become an infrastructure technology -- Providing cost effective edge infrastructure supporting MPLS services at the Edge and hicap transport infrastructure in the metro core.

But the notion of a ubiquitous and transparent Ethernet services is going to disappear, it never was aligned with how customer's build their networks anyways and never came close to providing the promised economies.
metroman 12/5/2012 | 2:57:33 AM
re: Ethernet Expo: Gurus Grade Themselves I think all of these are valid points but they do paint a slightly distorted and narrow view.

Taken in isolation it is true that Ethernet is not cheap - almost the only commodity components that exist in a Carrier grade ethernet device are a MAC chip and a physical port. Almost all other key components are custom built for purpose (if they are doing it right), this will drive up the price and the "cheapness" of Ethernet goes away. Add to this the complexity of MPLS control planes and you are no longer cheap. Alternatively, try deploying ethernet with Spanning Tree as a control plane and you will be in a world of pain. Ask many operators around the world who suffer today..

Early Ethernet services used enterprise Ethernet switches that are cheap for network infrastructure and the service could therefore be cheap. People want more than best effort services in any case.

SPs are never going to make money selling "Ethernet". They should be selling what Ethernet does. It does reduce the end to end cost and complesity of deployment by not requiring costly and complex interworking with other technologies. Any ethernet port in a router will do as a CPE in most cases, no need for OC3/STM1s or T3/E3s. For the operator, they can collapse other layers into Ethernet as long as the solution is reliable and scalable enough.

If you can then run many services (Voice, Data, Video, DR, Mirroring, gaming, etc) on one infrastructure then you will reduce cost and therefore be able to offer a premium service at a competitive rate while differentiating yourself and protecting tarrifs from erosion.

The only good thing about Ethernet in all of this is that it is the first transport architecture that allows true service based networks (not a network per service) without too many limitations. (there are some...)

Ethernet will not be for everyone, but it will be for most people a medium to start towards "one network" companies. This might frighten some people who survive in this industry by operators having multiple parrallel networks. Strong, creative and disruptive competitors will force the hand of the SPs sooner or later. Until there is a compelling event in North America this will not happen. Why cannibalise your reliable legacy services when you don't need to.. unless you want first mover advantage of course.

Ethernet may not end up being the nirvana technology to build "one network" operators, but it shows the way forward, and that it is possible.
metroman 12/5/2012 | 2:57:33 AM
re: Ethernet Expo: Gurus Grade Themselves one more thing eramk..

"I don't think its fair to blame the SPs. It was the vendors who introduced the market to the notion of Ethernet's "new service economies", "

this is pretty funny.... Do you do everything a vendor tells you? My advice would be to do your own due dilligence and not buy based on powerpoint.
Sign In