solving STP and VTP in DC enviornment

Unanswered Question
Feb 21st, 2007

hi,

i m having 2 65xx Core switches with me and around 10 49xx and 10 3560 switches with me ..connected in a triangle (65xx with etherchaneel and all rest of switches with dual connection to both the core switches.

need to know what is the best way to configure them so as to avoid STP loops and also for VLAN's flow .

shall i divide the switches in half and make both the core as root ...or one as root ...

Kindly Help...

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.

A broad question. In our case we use 2 Core Switches as VTP Servers. Odd VLANs on 1 switch, even on the other. The core switches are trunked (10Gige/Fiber)to each other with applicable MFSC cards configured with HSRP for fail-over.

Each distribution switch is dual connected to both Core Switches (1Gige Fiber)(1 connection to each Core Switch). We use RPVST for STP with default settings.

RPVST is snappy. Just try to limit Distribution-layer-to-Distribution-layer interconnects... that is where there might be a potential for STP. Same holds true for Access layer switches. Make everything go to each core switch and let the Core's take care of STP resiliency.

hoogen_82 Wed, 02/21/2007 - 23:23

Well my opinion would be to go ahead with your design. Keeping it simple is the key only configure MST or RPST if you know what your doing with the config. PVST+ is simple and enabled by default. And as you say it is a Datacentre, i believe your design would be final in these cases just go ahead with PVST+. Use VTP transparent mode on all your access switches. This way you can keep control of the vlan information passing to your access switches. The core can run on VTP Transparent too.

Redundancy is an important factor. As you have mentioned you are using your 6500 as core, just another question are you having a single sup or double sup engine?

For your HSRP do load sharing across your core switches odd vlans given higher priority in Primary Core and even vlans given higher priority in Secondary core. Similary do remember the spanning-tree priority command.

For example vlan 10 which is having higher HSRP priority on Core Switch 1 should also have spanning-priority lower i.e probably keep it as 8192 and this spanning priority for Vlan 10 on Switch 2 should be 16384. Load share your layer 2 also across switches.

HTH

Hoogen

Do rate if this helps :)

Actions

This Discussion