This evening I received an email forward containing the message below. It indicates that Fibre Channel over Ethernet (FCoE) has finally been standardized by T11.
From: Silvano Gai (sgai)
Sent: Thursday, June 04, 2009 5:32 PM
To: Silvano Gai (sgai)
Cc: Claudio DeSanti (cds); Landon Curt Noll (chongo); Joe Pelissier (jopeliss)
Subject: FCoE IS STANDARD - FOR PUBLIC RELEASE
***** FOR PUBLIC RELEASE – PLEASE DISTRIBUTE AS WIDELY AS POSSIBLE ****
“The FCoE standardization activity started in April 2007 is now successfully completed.
On Wednesday June 3rd, 2009, the FC-BB-5 working group of T11 completed the development of the draft standard and unanimously approved it as the final standard. The following day the plenary session of T11 approved forwarding the FC-BB-5 standard to INCITS for the publication process as an ANSI standard.
Cisco Systems has significantly contributed to the definition of FCoE in the FC-BB-5 standard through multiple technical submissions. Cisco has led the FC-BB-5 working group holding the chairman position with Claudio DeSanti.
One of the most significant technical challenges for the working group has been the definition of the addressing mode for FCoE, with two schemes proposed, FPMAs (Fabric Provided MAC Addresses, proposed by Cisco) and SPMAs (Server Provided MAC Addresses, proposed by Brocade). The final standard recognizes the superiority of the Cisco scheme by stating that FPMAs are the only addresses mandatory to implement.
Please join me in congratulating the Cisco T11 standards team (Claudio DeSanti, Joe Pelissier, and Landon Curt Noll) for this terrific achievement.
This information can be publicly released. The FCIA and Cisco plan to have press releases next week.
Wikipedia and www.fcoe.com will be updated during the weekend.
In October there will be a new edition of the FCoE book published by Cisco Press, updated to the standard.
Please distribute this note as widely as possible.
The standard is at:
http://www.t11.org/ftp/t11/pub/fc/bb-5/09-056v5.pdf
I just noticed that Ruv Cohen posted a similar article shortly after my post on the LISP working group. He includes some comments on the impact that LISP may have on Cloud Computing, and I think he’s got it right. Without going into great depth, it’s worth pointing out that a Locator/ID Separation enables a kind of mobility. Not mobility in the “mobile node” sense, such as a cellular phone. But, rather, coarse mobility of live networked devices between different datacenters.
The general idea of separating the host ID from the Locator is as simple as inserting an abstraction layer. Today an IP address is used to identify an endpoint on the network, such as an Ethernet interface on a server (like 153.16.10.6). It is also aggregated together into a subnet “prefix” (such as 153.16.0.0/16) and used to route traffic in the network backbone, thus acting as a Locator.
BGP routing table entry for 153.16.0.0/16
Paths: (1 available, best #1, table Default-IP-Routing-Table)
Not advertised to any peer 356117412803943
208.172.2.62 from 208.172.2.62 (208.172.2.62)
Origin IGP, localpref 100, valid, external, best
Community: 3561:3106 3561:21000
Last update: Sat May 9 20:13:52 2009
By inserting a protocol shim and “jacking-up” the endpoint ID we are able to decouple it from the Locator prefix. The ID still looks like an IP address, still fits into an IP packet header, and is compatible with the networking stack in existing host operating systems. But devices in the network called Ingress/Egress Tunnel Routers encapsulate the packet in another set of headers and forward it through the backbone based on an Endpoint ID (EID) to Locator mapping.
Map-cache entry for EID 153.16.10.6:
153.16.10.0/24, uptime: 00:00:01, expires: 23:59:58, via map-reply, auth
Locator Uptime State Priority/Weight Packets In/Out
128.223.156.134 00:00:01 up 1/100 0/0
The mapping is everything, and is currently the center of LISP research and experimentation. But a good mapping system would enable a host to move by simply changing the Locator mapping for the host’s EID. If for instance the host were a virtual machine running in the cloud, then it could theoretically move (motion or live migrate) between datacenters without dropping any existing connections.
Pretty cool! And definitely worth your attention if you’re interested in cloud networking.
I just received the following message, and thought I’d pass it along. I plan to attend the talk so if you’re going to be there please drop me a line and I’ll look for you.
The St. Louis Section of the IEEE Communications Society Presents:
Advances in Broadband Access Communications
Professor Tho Le-Ngoc, McGill University
Date: Wednesday May 6th Location: SAVVIS Communications
1 Savvis Parkway, Town & Country MO Time: 12:00pm-1:30pm
(Lunch Provided) Cost: FREE
Abstract: Broadband access communications have been continuously developed with many advances in response to increasing demands of multimedia services and applications. Broadband communications systems are generally power‐ and bandwidth‐limited while the communications channels are characterized by their frequency‐selective fading and the presence of noise and interference. Due to the dynamic nature of both transmission media and multimedia traffic, re‐configurable and adaptive schemes are of particular interest. Furthermore, in most broadband communications systems, interference is the major performance-limiting factor and interference mitigation can be more efficiently done by coordination of the transmission and resource allocation between users to minimize mutual interference in a collaborative manner. The talk will provide an overview of new developments/deployments in broadband access communications, and highlight technical issues in dynamic resource allocation and intelligent multi‐dimensional signal processing and the evolution from competition, to cooperation and cognition in communications.
