home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!gatech!usenet.ins.cwru.edu!slc6!trier
- From: trier@slc6.ins.cwru.edu (Stephen C. Trier)
- Newsgroups: comp.protocols.tcp-ip
- Subject: Re: TCP congestion/Nagle's algorithm
- Date: 25 Jan 1993 19:09:55 GMT
- Organization: Case Western Reserve University, Cleveland OH (USA)
- Lines: 24
- Message-ID: <1k1du3INNdoe@usenet.INS.CWRU.Edu>
- References: <1993Jan23.113544.5652@nmsu.edu>
- NNTP-Posting-Host: slc6.ins.cwru.edu
-
- In article <1993Jan23.113544.5652@nmsu.edu> amolitor@moink.nmsu.edu (Andrew Molitor) writes:
- > Second, Nagle seems to be saying that a TCP should not
- >send new data while unack'd data is outstanding, while Karn's
- >code seems to not send new data while unack'd data is outstanding
- >*unless* there is a max segment sized chunk to go out.
-
- I thought the latter is what Nagle said, too, but I just went and reread
- it, and I think you're right. He doesn't talk about sending immediately
- when one can send a full packet. RFC 1122 amends the 896, though, and
- 1122 does say that one should not wait for an ack before sending an MSS-
- sized packet.
-
- Perhaps what happened is Jacobson's Slow Start and Congestion Avoidance
- algorithms provided a better way to solve the broader congestion problem,
- so Nagle's algorithm is now used only for smaller-than-MSS packets.
-
- RFC 1122 has a nice description of how all of the different "when to send"
- algorithms, including Nagle's, fit together.
-
- --
- Stephen Trier "We want to offer you a price that you
- Network software type just can't afford to take advantage of."
- Case Western Reserve University - Sales blurb from HSC Software
- trier@ins.cwru.edu
-