Re: [nsp] BGP maximum datagram size

From: Stephen Sprunk (ssprunk@cisco.com)
Date: Tue Oct 30 2001 - 17:30:17 EST


Thus spake <jlewis@lewis.org>
> On Tue, 30 Oct 2001, Stephen Sprunk wrote:
> > Okay, TCP has for some reason negotiated a MSS of 536, which means
> > a maximum datagram size of 576, assuming you're using IPv4 and there
> > are no IP or TCP options present.
>
> Other than a small MTU somewhere on or between the routers, why might
> they do this? I just checked on a few routers (7513, 7206, 3640) running
> both iBGP and eBGP with peers at the other ends of point to point T3's,
> T1's or across our WAN. All of them show:
>
> Datagrams (max data segment is 536 bytes):

RFC1191 Section 3.1

"A host doing PMTU Discovery must obey the rule that it not send IP
datagrams larger than 576 octets unless it has permission from the receiver.
For TCP connections, this means that a host must not send datagrams larger
than 40 octets plus the Maximum Segment Size (MSS) sent by its peer."

"Some TCP implementations send an MSS option only if the destination host is
on a non-connected network. ... Actually, many TCP implementations always
send an MSS option, but set the value to 536 if the destination is
non-local."

You will likely see significantly faster BGP convergence if you use:

ip tcp selective-ack

ip tcp mss 1460

ip tcp window-size 65535

ip tcp queuemax 50

ip tcp path-mtu-discovery

S



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:53 EDT