From: Bob Chahal (bob.chahal@xxxxxxxxxxxx)
Date: Tue Aug 07 2001 - 19:11:57 GMT-3
In some practice labs I have done when asked to configure TR-dlsw-Etherent
you are expected to set largest frame on the local peer with the token ring
as 1500
R1
source-bridge ring 10
dlsw local peer 1.1.1.1 lf 1500
dlsw remote 0 tcp 1.1.1.2
or alternatively set the remote peer on the other router like
R2
dlsw bridge-group 1
dlsw local peer 1.1.1.2
dlsw remote 0 tcp 1.1.1.1 lf 1500
But is this really necessary. particulary if the routers serial interfaces
have a mtu of 1500 then a frame coming from a token ring device which may
have a frame size of say 4472 will get fragmented anyway or is this
something entirely different. If you read the design guide it also suggests
that when using TCP encapsulation you do not need to set the LF parameter.
It goes on to say
"if the LF is smaller than the MTU, then TCP will fragment each packet and
send them in sections across the WAN. If the LF is larger than the MTU, then
individual packets will be placed into the TCP/IP frame."
Surely that is the other way around.
Any thoughts would be appreciated.
Thanks
Bob
turning on path MTU discovery. Whilst on the subject of the DLSW design
Guide it even has this statement which to me seems totally wrong.
**Please read:http://www.groupstudy.com/list/posting.html
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:31:46 GMT-3