Wired Intelligent Edge

Hi experts,

 

Basically I have a topology where I have a stack of two 3810M switches as core, and I am going to install two access switches in every floor. I know the best option is to have a VSF of two 2930F access switches and deploy a LAG between each VSF in every floor and the core. But because money is a constraint, I wonder if I can use two 2540 as access switches and deploy a distributed trunking between each pair of 2540 switches in every floor and the core, since the 2540 datasheet says:

 

IEEE 802.3ad link-aggregation-control protocol (LACP)
and port trunking support up to 26 static, dynamic, or
distributed trunks with each trunk having up to eight links
(ports) per static trunk.

 

Thanks in advance,

Julián

Greetings!

 

This appears to be a typo, as distributed trunking is only supported on the 3810 and 5400R zl2 series switches (as well as the previous-generation 3500, 3800, and 5400 zl switch series).

 

The 2540 (and any other Aruba switch) supports connections to a distributed trunking pair using LACP or static trunking (which is possibly what is being implied by the quoted statement), but does not support DT on its own.

Hi Matthew,

 

Thanks for the clarification, so my best option is to install VSF with 2930F and deploy LAG to the stack core. If I install 2530/2540 they can't do stack and DT and I will have to use STP between them and the core.

 

One more doubt, for the 3810 and 5400 switches, is there any advantage of using DT when those switches support stacking? AFAIK, DT sees a LAG distributed between a pair of switches as if the links comes from the same logical switch (which is also true with backplane stacking and VSF), but the configuration is more complex and the two switches are still managed independently. So I don't see the point of using DT instead of forming a stack (backplane or VSF) and configuring a normal LAG with ports on both switches.

 

Regards,

Julián

Distributed trunking would be used as an alternative to backplane/VSF stacking in cases where you would want to maintain separate control planes between the two members. This enables use cases requiring taking one of the two switches down for maintenance or software upgrades without affecting the remaining switch.

 

The tradeoff of distributed trunking is having two separate devices to manage and monitor, where a stack presents a single logical device with a single point of management.

Hi Matthew,

 

OK, thanks very much for the clarification!

 

Regards,

Julián

Hello Julian, also consider that, historically, DT was available years before VSF introduction (which is recent)...so the tradeoff was simply that it was (especially for 5400 zl and 5400R zl2) the only specifically designed method of aggregating two chassis through ISL+Keep Alive links and letting downlink switches to connect to both chassis via a normal LACP (or via to a non Protocol) Port Trunk. There were simply no other options.