[c-nsp] jumbo frames

[Tim Winders]

Lots to catch up on...  Thanks for all the discussion.

OK, this is what I understand thus far:

There is no reason not to set the maximum MTU size on all L2 interfaces.
An increased MTU should only be set on an L3 interface if all hosts on that
network support the larger MTU size.
In practice, to take advantage of a larger MTU your hosts should be on the
same network.

So, in testing this out, I have set the MTU size on my L2 interfaces and
globally on my 3560G-GE's to 9216 and 9198 respectively.  I have let the
default MTU on the L3 VLAN interfaces.  I have set my Linux hosts to have an
MTU of 9000 and my Windows hosts to have an MTU of 9014.  The hosts are all
on the same subnet.

How can I test to make sure the larger MTU is actually being used between
hosts?

When I do a tracepath from the Linux hosts I see pmtu responses for 9000,
8166, 4352, 2002, and 1492.  Then at the bottom it reports "Resume: pmtu
1492 hops 1 back 1".

I can do a "ping -s 8500 -M do ipaddress" and I get a response.  So, it
seems like things are working correctly, but the pmtu of 1492 above is
bothersome to me.

---

Tim Winders
Associate Dean of Information Technology
South Plains College
Levelland, TX 79336 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3106 bytes
Desc: not available
Url : https://puck.nether.net/pipermail/cisco-nsp/attachments/20060304/25a88486/attachment.bin