cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
570
Views
0
Helpful
1
Replies

Blade switch STP advice

valsidalv
Level 1
Level 1

Hi,

I'm configuring environment according to attached picture. In this infrastrucure will be blade servers with teaming NFT, so only one network card will be up.

Could you please advice me the best spanning tree design for this infratructure?

Thank you very much.

Best regards,

                       Vladislav

1 Accepted Solution

Accepted Solutions

gnijs
Level 4
Level 4

If your cores are by any change N5Ks or N7K of C6500/VSS , then go with vPC and MEC portchannels, eliminating STP loops completely.

If not, I wouldn't dual uplink my blade switches in this design. I would connect Blade 1 with Core 1 and Blade 2 with Core 2.

Then use "link state tracking" for redundancy and failover on the blade switches.

Therefore, no spanning tree loops and 50% less ports used on my core & blade switches.

There is only one small disadvantage with this design: if you have silent devices as servers (ie some linux servers) and your server doesn't broadcast a lot, on switchover to the other blade & core switch, the MAC cache might not get updated immediatly. This might be solved by the blade switches itself (ie HP VC switches) or by the NIC teaming drivers (ie. gratuitious ARP on switchover) or by using "heartbeat hello" on the NIC teaming configuration (everything that generates traffic is fine :-)

regards,

Geert

View solution in original post

1 Reply 1

gnijs
Level 4
Level 4

If your cores are by any change N5Ks or N7K of C6500/VSS , then go with vPC and MEC portchannels, eliminating STP loops completely.

If not, I wouldn't dual uplink my blade switches in this design. I would connect Blade 1 with Core 1 and Blade 2 with Core 2.

Then use "link state tracking" for redundancy and failover on the blade switches.

Therefore, no spanning tree loops and 50% less ports used on my core & blade switches.

There is only one small disadvantage with this design: if you have silent devices as servers (ie some linux servers) and your server doesn't broadcast a lot, on switchover to the other blade & core switch, the MAC cache might not get updated immediatly. This might be solved by the blade switches itself (ie HP VC switches) or by the NIC teaming drivers (ie. gratuitious ARP on switchover) or by using "heartbeat hello" on the NIC teaming configuration (everything that generates traffic is fine :-)

regards,

Geert

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Innovations in Cisco Full Stack Observability - A new webinar from Cisco