Link Aggregation vs Spanning Tree

[ITGURU]

Hi,

I've always ran a single link for core switches to distribution switches and never had a failure. However, to further improve redundancy, I want to configure 2 links between edge and distribution for redundancy.

 

On the core switch i have 4 switches in stack configuration, and can Link aggregate across multiple switches.

However, on the distribution switch I can only link aggregate ports on the same chassis switch slot.

 

Therefore, i think i have 2 options:

 

Use LACP for the 2 links between core and distribution switches but this will only cover a SFP module failure on the same slot, and the other port remain active. If the chassis line card failed both link would fail.

 

Therefore if i connected 2 links from the core stack (2 different switches) to 2 different line cards on the distribution chassis would i then enable STP?

 

I've not setup STP before but believe the core switch needs to be set to a value of 0 and then i need to set ALL non redundant link ports to edge?

 

If anyone can help/advise so i can get this implemented on Monday, that would be great.

Thanks

[FN-GM]

Hello,

 

You would need STP. But this should be setup already to stop loops? If not you need to consider what version of STP you use. Some versions can be horrifically slow this combined with some switches and certain configuration can mean clients connecting to the network take ages to get IP addresses.

 

The value to spoke about is the root bridge. I will avoid setting this as 0. I would pick something higher. The reason being is if you need to replace the core in the future and have both old and new running side by side you can’t make the new core the root bridge as you can’t go below 0. It will work fine, but will remove flexibility for the future.

 

What type of switches are you using? If you use Cisco it can be configured with just a few commands and you’re good to go. Others might need some more consideration.

 

Thanks.

[ITGURU]

Hello,

 

You would need STP. But this should be setup already to stop loops? If not you need to consider what version of STP you use. Some versions can be horrifically slow this combined with some switches and certain configuration can mean clients connecting to the network take ages to get IP addresses.

 

The value to spoke about is the root bridge. I will avoid setting this as 0. I would pick something higher. The reason being is if you need to replace the core in the future and have both old and new running side by side you can’t make the new core the root bridge as you can’t go below 0. It will work fine, but will remove flexibility for the future.

 

What type of switches are you using? If you use Cisco it can be configured with just a few commands and you’re good to go. Others might need some more consideration.

 

Thanks.

I have dlink switches and was looking at RSTP.

 

I don't have spanning tree enabled because the switches have loop back detection enabled across all ports so of any ports get looped it will disable both and send an snmp notification to me and everything I running as single uplink from switches at the moment so stp isn't required to stop loops

[snagrat]

I have dlink switches and was looking at RSTP.

 

I don't have spanning tree enabled because the switches have loop back detection enabled across all ports so of any ports get looped it will disable both and send an snmp notification to me and everything I running as single uplink from switches at the moment so stp isn't required to stop loops

 

Did you enable RSTP on your flunk switches in the end? And if so did you just accept the default settings?

 

I’m looking to do the same but wonder if I need to configure them in a specific way. We have dlink at Core and Edge.

 

And finally, did you keep loop back detection turned on for all ports of you did enable RSTP?