Probably more to do with Novell. Check that the customer is running client
shells and a version of the Server OS that support pburst. Otherwise,
every NCP packet needs to be ACKed. Given your line speed and the double
encapsulation (frame, tunnel, novell), you're results aren't really that
far out of whack for the ACK-per-packet old-shell behavior.
You may also want to turn on whatever you can on the Cisco's to eliminate
or tone down the broadcasting associated with SAP/RIP/Watchdog, etc. and
save yourself some bandwidth. Guides to doing this are in our documentation.
At 03:54 PM 1/16/98 +0200, you wrote:
>Hello all,
>
>we have the following configuration for one of our customers:
>
>SiteA ------- GW1--------GW2------SiteB
>
>Our GWs are Ciscos while the routers at the customer's sites are 3com,
>therefore :-(
>we use Frame Relay between SiteA<->GW1 and SiteB<->GW2
>and a tunnel so that SiteA and SiteB are "directly connected".
>
>The customer connects two LANs that also run Novell.
>When they try to copy a file they get as much as 8Kbps at the most,
>even if there is available capacity along the path (33.6 Kbps customer links,
>128Kbps GW1-GW2 link used at 40%).
>
>Sh int tunnel0
>gives a 9Kbps speed by default, is this true?
>Can we alter that by changing the delay parameter of the tunnel interface
> or do you think it has to do with Novell?
>
>Thanks in advance
>Dimitris Zisiadis
>FORTHnet
>
>
>
>
>
>
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:13:14 EDT