Technology, Innovation & Telecomm

I’ve found it kind of interesting that for the last several years, “Carrier Class Ethernet” has been a tag line for various products. The notion behind that phrase is that traditional LAN-based Ethernet doesn’t have the reliability and quality of service (QoS) that other transports like ATM, SONET, SDH, etc. have always had. All very true.

So, various vendors and industry consortia have been trying to get it right. Standards have emerged that are intended to address the gap and to make it possible for Ethernet to be the true transport for the world’s IP and other traffic. MPLS was developed, for example, to address a few needs. One was the need to make it possible to establish a known route through a network, between various end-points. Another reason was to allow equipment to be able to make switching decisions very rapidly, doing the least amount of frame disassembly and inspection as possible.

All that is to say that Ethernet has had kind of a rough road becoming the primary delivery vehicle throughout a Carrier’s network. SuperComm this year has seemed to focus more heavily than ever on IP-based services (voice, video, data, etc. - see the prior post), with a tacit belief that all this will eventually be carried by good old Ethernet.

Here’s some interesting information (based on various vendor’s opinions):

1. There needs to be better QoS features within Ethernet. Vendors then need to make consistent use of them so that end-to-end quality can be assured.

2. Ethernet is quite complex to set up and run when it lives within the core of a large Metro- or Wide- area network. Don’t presume that it’s somehow vastly easier than current technologies.

3. While ATM is assuredly on the decline, last year more ATM gear shipped than ever before. Vendors have de-emphasized development of ATM gear in favor of Ethernet, but these things take time for adoption, once it starts.

4. To achieve the kinds of bandwidth and bandwidth density that the market feels will be necessary, a lot of the existing SONET/SDH infrastructure may not be adequate. If you believe that a consumer will need 50 Mbps, that seems very true. Thus, carriers are pushing fiber directly to xDSL DSLAMs at Gig-E and 10Gig-E line rates. That’s expensive and time consuming.

5. It’s not clear to me whether Ethernet in the core is really any cheaper for the carriers or the consumers directly. I think that various vendors are in the same predicament. Time will shake a lot of that out, but at the moment it could well be mainly customer perception, rather than reality, that shapes the desire for “All Ethernet, All the Time”.

That’s all the news that’s fit to print and I have to get ready for another day at the show.