From: Andy mrozek (andymrozek@yahoo.com)
Date: Fri Aug 06 2004 - 18:44:01 GMT-3
Request for opinions thoughts........... In a "campus
design" where you have say core SW1 and core SW2 at
the hub of the network which houses main centralized
data centers and majority of network resources... What
is the preferred method of connecting core Sw1 and
core Sw2 ....... I have noticed the following :
SW1 = core Fast0/1 and Fast0/2 are trunked and
channeled to the same ports on core SW2 represented by
the 2 dotted lines between in diagram below....
SW3 = represents a physically remote site attatched by
x2 single mode fiber runs say over a 6 mile distance
within a city......
All switches are within the same vtp domain so all
vlans are same throughout domain..... If we do
intervlan routing on SW1 / SW2 say via a sup engine 4
or 5 in a catalyst box and run HSRP for redundant
default gateways for all the remote sites and say in
the below diagram SW3 is root for Vlan 50 which is
permitted across all trunk links , so that means port
channel between SW1 and SW2 core's will block as they
will always have there root port , which in this case
SW3 in a forwarding state..... My observations are
via sniffing / debugging that the hsrp communication
is obviously going to pass between SW1/SW2 through SW3
.... I don't see of a way to avoid that situation...
and it would seem that this method of campus design
would not be scalable ... What if we had say 1000
remote sites hence 1000 + SW3's ..... trunking up to
the core ......... even with a well thought out vlan
pruning we would have all this multicast hsrp from
224.0.0.2 shooting throuout the network between the
core switches...
Any thoughts ??? Thanks Andy
SW1 ------------- SW2
-------------
| |
| |
| |
| |
\ /
\ /
\ /
\ /
SW3
=====
-Andy Mrozek
There are only 10 types of people in this world...
those who understand binary, and those who don't.....................
This archive was generated by hypermail 2.1.4 : Fri Sep 03 2004 - 07:02:34 GMT-3