At 6:04 PM +0000 8/13/04, Priscilla Oppenheimer wrote:
>e the core of the network?? Just kidding.
>
>> I
>> keep
>> coming back, in horror, to running two GRE tunnels, one from
>> each
>> 3550 to each distribution 6509. If there's some way to make the
>> uplink act as both a routable subnet while simultaneously
>
>I don't think this tunneling idea is as ugly as it sounds. As you mentioned
>earlier, you'll basically have this large bridged network floating in space,
>connected via the routed network. You'll be doing bridging over IP. It's not
>as bad as it sounds and there is precedence for things like this. It's a bit
>like the old remote source-route bridged networks that used IP for
>transport. What goes around comes around, so to speak.
Remember that one of the reasons we moved from RSRB to DLSW+ was that
DLSW+ had an assortment of scalability techniques -- hierarchy,
on-demand, etc., to reduce the processor load from all the TCP
tunnels. These 6509's have Supervisor 1A's, and I'm concerned about
processor load. It's something I want to watch very carefully.
While the 6500 is better balanced between forwarding and processing
than was the 7000, I remember several SNA appliations where I'd
front-end a channel-attached 3-available-slot 7000 with one or more
4500/4700s. The SNA tunnels terminated on the RISC 4x00s, which had
the CPU power to handle lots of tunnels. I then routed over multiple
FDDI to the 7000, which had the bus performance to handle the
forwarding to the mainframe through the CIP. The poor little 68040 in
the 7000 could never have handled the tunnels.
Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=91761&t=91736
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
