10G Network Backbone redesign recommendations

[RobB_8]

I am interested in recommendations... I am considering redesigning the 10G network backbone in our main site to improve resiliency. But because this Lanroom is the primary hub for our Internet, WAN link to 58 sites, and all our central servers. I want to be 110% sure before attempting this. Currently we have a 6400 switch with 5 3400 swtiches atttached via 10G tranceivers in a star. 3 3400s are for the main lanroom and are connected via 10g copper tranceivers. The other 2 3400's are mid tier switches connected to the secondary closets via 10G fibre. I have 2 extra 10G fibre tranceivers, and was considering the benefits of creating a ring for the 3 lanroom switches to the 6400.
I have some concerns about mixing copper and fibre 10G modules in the ring.
I am also undecided as to the best option for creating the ring for load balancing, fault tolerance, traffic patterns, recalculation times, switch performance etc.
I am interested in any suggestions.
Thanks

[Mohieddin Kharnoub]

Hi Rob

It sounds interesting idea, and i agree with you in your thoughts to keep a resiliency in each group of switches.

I believe in redundancy, and in everything, and in your case i believe redudancy can be done based one 4 things mainly:

- Resilient links for each IDF (group of switches) by the use of 10Gig.
- At least 2 10 Gig links (4 will be better if you have a dual core) from each switch group to the Core.
- L2 redudancy by the use of MSTP, and load balance between the Vlans over the redundant link from edge to core.
- L3 redudancy by the use of XRRP if possible.

The 10G FAQs has some good hints:
http://www.hp.com/rnd/support/faqs/10-GbE-trans.htm

Switching Performance:
The 3400-48G has 136 Gbps switching capacity, where the 6400 has 160 Gbps and thats why its considered an Aggregator.

Good Luck !!!