Re: [RE] Preliminary report, 802.1 liason
At 05:48 13/05/2005 -0700, Richard Brand wrote:
>John:
>While I agree with your statement in principal I trust that you are not
>proposing "Res A". No consumer company that I know will swallow that
>and I have to believe that you can't either.
>What you do highlight though is that we absolutely need is changes to
>existing 802.3 to support these new home focused applications.
>FYI, I am sitting right now in the DSL Forum meeting where the topic is
>DSL Ethernet migration. At this moment, heavily deployed ATM carrier
>Deutsch Telecom is describing how they are moving to EtherSLAMs. 5
>years from now, there will be little ATM left in the broadband access
>connections to residents world wide.
That's what I find so bizarre. ATM Forum developed a system with all the features that are right for media streams, and then positioned it as a data network which didn't use most of those features and needed other features to be added (I'm thinking of ABR). Now we have increasing demand for media stream traffic, it all goes over data networks which are singularly inappropriate for it and need all kinds of kludges to make it work. And ATM Forum's contribution has been to bring out a standard (whoever called it FATE had a sense of humour) which (a) supported AAL5 only (enshrining the idea that it's a data network) and (b) encouraged a retreat from one of the few areas where their technology had a foothold.
MPLS could be viewed as the result of applying ATM-like technology to solve a problem in the core network. In the same way, ResE may well find that an ATM-like technology would be the best way to meet its requirements. The words "re-invent" and "wheel" come to mind.
I don't think the non-availability of 1990s-style ATM services in the wide area will be a problem; the home network will also need to interwork with a number of other media, such as satellite and digital broadcasting, so should be able to connect to whatever is flavour of the month telecom-wise by then.
>Therefore who do you see will need ATMoE to the residence?
I don't see "to the residence" as being particularly important at this stage. I do see ATMoE as lower complexity than RTP-over-UDP-over-IP-over-Ethernet for media streams, particularly if you do some simplification like saying you'll only support CBR multicast and UBR point-to-point.
John Grant
___ ___ ___ ___ ___ ___ ___ ___ ___
| || || || | | || || || || |
| N || i || n || e | | T || i || l || e || s |
|___||___||___||___| |___||___||___||___||___|
Nine Tiles Networks Ltd, Cambridge, England
+44 1223 862599 and +44 1223 511455
http://www.ninetiles.com