I was visiting a medium-sized ISP with lots of IETF people today, and an
old issue came up: default TCP MSS in NetBSD.

The Ethernet MTU has been the standard MTU for DS0 (64Kb/s) and T1 leased
lines in cisco equipment for a long time, and they're still the majority of
the backbone last I checked. MTU for FDDI and T3 is even higher.

I realize that net.inet.tcp.mssdflt is now a sysctl variable that can be
set per system; I'm suggesting that perhaps our default ought to more
closely reflect the reality of modern IP networks. I think it would make
everyone happier to be shooting bigger packets into the networks that
NetBSD is attached to - the ISPs, and the customers.

It strikes me that there are two ways this can be done:

1. change sys/netinet/tcp.h TCP_MSS to 1460

2. add a line to /etc/netstart:
	/usr/sbin/sysctl -w 'net.inet.tcp.mssdflt=1460'

I'd prefer to see #1, of course.

What are the real (not theoretical) downsides of this change? I'd like to
see this go in to NetBSD 1.3. More througput with lower overhead.

	Erik <fair@clock.org>