* Grant <emailgr...@gmail.com> [160920 08:53]:
> >> > A while back I was having networking issues. I eventually tried
> >> > drastically lowering the MTU of all the systems onsite and the issues
> >> > disappeared. I always thought the issue was due to the MTU on our
> >> > modem/router. Today I read that AT&T DSL requires a 1492 MTU so I
> >> > increased the MTU of our systems up to 1492 and haven't had any
> >> > issues. Do certain ISPs require you to change the MTU of your entire
> >> > network, or is this likely due to our AT&T modem/router itself?
> >> AFAIK the MTU is defined for every network interface separately. For an
> >> ADSL connection it is common that a lower MTU is needed because of the
> >> PPPoE header information that is encapsulated in the ethernet frames.
> >> But in that case it is sufficient to lower the MTU just for the WAN
> >> interface that is connected to the DSL modem.
> >> If you don't use protocol encapsulation in your LAN then there should
> >> be IMHO no reason for lowering the MTU of your internal interfaces.
> >> --
> >> Regards
> >> wabe
> > MTU is per network interface but you really don't want to end up having
> > your router fragment every IP packet because systems on your subnet are
> > using a larger MTU.
> > Todd
> That makes sense. So in my case, I'm thinking 1492 MTU on every
> interface in the network.
> So I'm sure I understand, should everyone with a DSL connection set an
> MTU of 1492 (or potentially lower) on all of their network interfaces
> to avoid packet fragmentation?
> - Grant
I would probably set the MTU to 1492 on each interface myself.
But it really depends upon the traffic mix and how "smart" ("dumb") the
devices on the network are.
TCP is likely using Path MTU Discovery to determine the TCP Maximum
Segment Size so that TCP traffic doesn't encounter IP fragmentation
end-to-end. PMTU used to use ICMP packets to determine the end-to-end
MTU, but RFC4281 (I think) uses a method to work around the dropping or
filtering of ICMP packets.
However, there's different quality of implementations as well as
differences in whether the network stack uses the PMTUD for UDP and
other datagram traffic or not as well.
DISCLAIMER: It's been a number of years since I've been involved in
implementing any IP networking so things may have moved on since then.