[c-nsp] Cisco IIH padding

[Mark Tinka]

On Tuesday, November 25, 2014 11:08:53 PM Vitkovský Adam 
wrote:

> Interesting I never heard of the "always" option one
> learns something new every day. Though there's still
> this IIH padding madness when it gets to the actual size
> of the IIH datagram and the allowed offset between CLNS
> MTU and the actual IIH size.

		An operational example:

We use "clns mtu" where there exist physical limitations on 
the link or the port.

In our case, CSR1000v runs on servers whose NIC's support 
9,000 bytes maximum. Our network runs 9,192 bytes in the 
core. This causes an MTU mismatch between the CSR1000v and 
the rest of the network.

The solution - physical MTU on the CSR1000v is set to 9,000 
bytes (maximum supported), but "clns mtu 8000" is used to 
get the initial padding through and have the IS-IS 
adjacencies form.

To avoid packet drops/blocks of BGP update messages, TCP MSS 
is set to 1,500 bytes on the CSR1000v. This ensures BGP 
messages do not get stuck in Adj-RIB-Out queues when transit 
MTU between the CSR1000v and the rest of the network does 
not match.

Mark.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <https://puck.nether.net/pipermail/cisco-nsp/attachments/20141126/eac7e237/attachment.sig>