Hi,
On Sat, Jul 07, 2001 at 11:07:03PM +0100, Ryan O'Connell wrote:
> Good L2 design is the key to avaoiding STP problems - "Cisco LAN Switching"
> has a couple of very good chapters on this.
There is just No Way you can do things like this:
A
/ \
B---C
with three switches A, B, C and many routers or other high-bandwidth
devices connected to all of them (think "IP exchange point" scenario) and
have packets flow from each switch to each other *directly*.
In any STP based scenario, one of the links will be STP blocked, leading
to packets travelling from (say) B to C via B->A->C. Which is BAD if you
have high-volume traffic going B->C, which *IS* going to hit you in
certain situations, no matter how good the network design - it usually
hits in ISP / exchange point applications much sooner as in the standard
"datacenter" stuff with a few high-bandwidth servers, connected to "The
Core Switch(es)" and lots of "smallish PCs" connected to wiring closet
switches having GigE uplinks to The Core Switches, but hardly any traffic
worth mentioning between different wiring closets.
Also, for one-router-many-ethernet redundancy, it would be very desireable
to be able to connect a router to both switch A and B in the scenario
above, and use the "best" path for your packets. What's available right
now just sucks in this setup.
gert
--
USENET is *not* the non-clickable part of WWW!
//www.muc.de/~gert/
Gert Doering - Munich, Germany gert@greenie.muc.de
fax: +49-89-35655025 gert.doering@physik.tu-muenchen.de
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:44 EDT