Date: Fri, 19 Jun 1998 23:44:51 -0700 From: bmah@CA.Sandia.GOV (Bruce A. Mah) To: spork <spork@super-g.com> Cc: Joel Ray Holveck <joelh@gnu.org>, opsys@mail.webspan.net, root@bmccane.maxbaud.net, hackers@FreeBSD.ORG Subject: Re: TweakDUN Message-ID: <199806200644.XAA24111@stennis.ca.sandia.gov> In-Reply-To: Your message of "Fri, 19 Jun 1998 22:17:09 EDT." <Pine.BSF.3.96.980619220851.25847A-100000@super-g.inch.com>
next in thread | previous in thread | raw e-mail | index | archive | help
[-- Attachment #1 --]
If memory serves me right, spork wrote:
> modem-> dialup PPP 1500 -> term server -> ethernet 1500 -> router -> T1(s)
> HDLC 1500 -> core router -> fast ethernet 1500 -> upstream's border
> router -> FDDI 40?? -> upstream core router -> ATM/SONET/whatever ?
FDDI: 4192?
ATM: 9188 (usually)
I don't use dialups very often nowadays, but I dimly remember trying to
negotiate a *smaller* MTU on a downlink, in order to try to get better
interactive performance (mumble mumble, use IP TOS bits and smarter queueing,
mumble mumble).
> Generally, one avoids small MTUs on big links, I beleive. ATM's small
> cell size makes *every* packet get fragmented at layer 2, but I'm not sure
> that's even relevant.
You're right, it's not relevent. :-) The IP layer doesn't see the
segmentation-and-reassembly to/from ATM cells (thank goodness...think of
having a 48-byte MTU). The ATM MTU of ~9KB was picked to be the same as that
of IP over SMDS.
> Anyone else? I've never heard of the oft quoted "Internet standard MTU of
> 576"...
RFC 1122 says that Internet hosts need to be able to *reassemble* a packet of
*at least* 576 octets, but that if you're sending data and you don't have any
other MTU information, you should assume an MTU of at most 576. ("Other MTU
information" includes knowing the link MTU if you're sending to a
directly-connected host, Path MTU Discovery results, and so on.)
This discussion reminds me of an ISP that I once consulted for. The users
kept doing various tweaks to Trumpet WinSock's PPP driver, including setting
odd-valued MTUs and TCP MSSs that had no relation to the MTUs. Unfortunately,
a lot of these folks didn't really know what they were doing. The results of
these tunings were pretty bizarre (TCP/IP Illustrated, Volume 3 has a chapter
entitled "Packets Arriving at a Web Server", or some such thing, which gives
some similar results).
Cheers,
Bruce.
[-- Attachment #2 --]
-----BEGIN PGP MESSAGE-----
Version: 2.6.2
iQCVAwUBNYtaYqjOOi0j7CY9AQEl9wP+PEncErdUExEVwtMAx2xhg08y931MvAkd
FlgndVg+pdPmck3+4kHipaSAXHTcJSHi8nxZ0967CQgo9y1HUFgh3k0XQcbPbNvk
jW+OV8mn2CBrzxJoa6bw9JS50M2Gg3hzWdyXs8FZfju9v+IHQ9J1/2m4h+rUpN8J
xKvDxvJXx+k=
=Pkaw
-----END PGP MESSAGE-----
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199806200644.XAA24111>