home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.dcom.sys.cisco
- Path: sparky!uunet!spool.mu.edu!agate!boulder!recnews
- From: dave o'leary <doleary@cisco.com>
- Subject: FECN and BECN bits on Frame Relay
- Message-ID: <724556611.16854@news.Colorado.EDU>
- Sender: news
- Date: 16 Dec 92 13:59:31 -0800
- Approved: news
- X-Note1: mail msgid was <199212162202.AA07231@spot.Colorado.EDU>
- X-Note2: message-id generated by recnews
- Lines: 52
-
-
- In article <723936986.11433@news.Colorado.EDU> Tom_Karolyshyn@dgc.ceo.dg.com writes:
- >
- (various header stuff deleted)
- >
- > Promotion of FECN Bits for OSI and DECnet Phase IV
- > Frame Relay switches have the capability of setting congestion bits
- > in packets (Forward Explicit Congestion Notification (FECN__ as they
- > transit the Frame Relay Network.
- >
- > This feature allows promotion of FECN bits from the Frame Relay
- > network to the appropriate congestion management fields of OSI and
- > DECnet Phase IV packets. The protocols are expected to recognize the
- > fields and provide some congestion relief by whatever mechanism that
- > is available to them. There are no new commands associated with this
- > feature.
- >
- >Do I read this correctly to mean that there is no benefit to TCP/IP
- >users, i.e., FECN bits have no use in a Frame-relay / TCP/IP
- >environment (like ours) ?
- >Tom Karolyshyn
- >Data General Corp
- >Westboro, MA 01580 508-870-7936
- >
-
- Since IP, TCP, UDP, etc. have no "congestion control" bits in their
- headers (like OSI and DECnet), there isn't anything the cisco can
- really do. I kind of wonder about the real value of this mechanism
- anyway, since the switch is setting the FECN bits because of
- congestion on an interface which may or may not be due to a specific
- host (and protocol and session on that host), i.e. some Appletalk node
- starts blasting away at another node across the frame relay through
- the cisco routers. A pair on DECnet hosts using the same set of links
- could receive packets with the congestion bit set because the frame
- relay switch sees too much stuff coming at it from the cisco router.
- Most of it might be from the Appletalk node (which will never receive
- any noticification of the FECN, directly or indirectly), and the frame
- relay switch can't distinguish between the Appletalk node and the
- DECnet node - it just sees stuff coming from the cisco. The DECnet
- node might back off but if the Appletalk node has more stuff to send
- it will just hog whatever bandwidth the DECnet session has given up.
- So the real problem is figuring out how to notify the offender that it
- is causing a problem in a multiprotocol environment with different (or
- no) means of congestion control per protocol and session. Trying to
- perform flow and congestion control at multiple layers in the stack
- that don't know what the other layer is doing (wouldn't want to violate
- that model :-) can cause unexpected results, in particular really bad
- performance.
-
- dave o'leary consulting engineer cisco systems
- (714)261-5681 Newport Beach, CA doleary@cisco.com
-
-