Wired Intelligent Edge

Hi, maybe somebody can explain the difference between the commands -> trunk a13 Trk1 dt-trunk (= HP Trunk) and trunk a13 Trk1 dt-lacp <- to me?

thank you!

You are referring to Distributed Trunking (HP/HPE/Aruba aggregation technology well described here): so dt-trunk and dt-lacp options are like, respectively, trunk and lacp options for trunk command (where trunk means aggregate two or more physical links together) BUT those options (dt-trunk and dt-lacp) refer to a Distributed Trunking scenario.

Do not confuse trunk with the meaning Cisco gave to the word (basically tagging more VLAN IDs on the same interface).

Edit: for the very same reason that dt-trunk option is different by dt-lacp option (exactly like trunk option is not like lacp option) I suspect the Switch 4 (top Access Switch 2520 in the presentation linked above, page 45 of 60) configuration line:

Switch4(config)# trunk 3-4 trk4 lacp

is wrong, it should have been instead:

Switch4(config)# trunk 3-4 trk4 trunk

The example (top Access Switch versus bottom Access Switch, both uplinked to a DT depolyment of two HP 5412zl switches) was provided to highlight the fact you can have Non Protocol Trunks and IEEE 802.3ad (LACP) Trunks that are terminating into the DT deployment.

Hi Parnassus,

Thanks for this information. I have read the 60-page presentation document. The DT technology makes the 2 DT switches act somewhat similar to VSX-peers. Is this another way of explaining how VSX works? I am just trying to "hammer" the difference between having 2 separate switches running DT to a single/stacked access switch, and having 2 VSX-peer switches running MC-LAG to a single/stacked access switch. In my work environment, I run MC-LAG on 2 ArubaOS 8320CX VSX-peer switches to 2 ports (1/A1, 8/A1) on an 8-stacked ArubaOS 3810M switch (as access switch). These ports (1/A1 & 8/A1) form a trunk group, trk1, running "dt-lacp". 

trunk 1/A1,8/A1 trk1 dt-lacp

I just want to know if I am doing the right thing by running "dt-lacp" on trk1 interface and not just "lacp". Or should I change it to just "lacp" since the VSX best practice guide actually used "lacp" instead of "dt-lacp".

Would appreciate a response.

Thanks.

---

@D.I. wrote: The DT technology makes the 2 DT switches act somewhat similar to VSX-peers.

I think DT and VSX can't be directly compared...eventually they can from standpoints of the peers (and not for sure in a like for like comparison): VSX is, from that regard and as I already wrote time ago, DT with steroids...but also VSX is running on a completely different OS with totally different software engineering history and new features. VSX offers much more than DT (or DT + VRRP).

---

@D.I. wrote: In my work environment, I run MC-LAG on 2 ArubaOS 8320CX VSX-peer switches to 2 ports (1/A1, 8/A1) on an 8-stacked ArubaOS 3810M switch (as access switch). These ports (1/A1 & 8/A1) form a trunk group, trk1, running "dt-lacp". 

 

trunk 1/A1,8/A1 trk1 dt-lacp

 

I just want to know if I am doing the right thing by running "dt-lacp" on trk1 interface and not just "lacp".

A stack of Aruba 3810M is necessarily deployed by using Hardware stacking modules and stacking cables (isn't it?)...DT is simply not in the picture...because you have a stack (virtual switch)...so the answer to your question above is: use simply Port Trunking (using LACP) on the Aruba 3810 Stack side...and corresponding VSX LAG (using LACP) on VSX side.

Thanks Parnassus.

Greetings!

Distributed trunking on the 5400, 5400R, 3800, and 3810 switch families is similar to VSX on AOS-CX in that you may distribute link aggregation across a pair of switches with separate control planes, with trunk state synchronization and inter-switch traffic forwarding between DT members; in the case of the 3800 and 3810, this can include a pair of switch stacks as well as standalone switches.

If you are only distributing links across members of the same stack, however, you would not use dt-trunk or dt-lacp, but rather the 'normal' trunk or lacp protocols when defining trunks. The dt-trunk and dt-lacp protocols are intended only for distributed trunking setups, where the trunk member ports are distributed across a distributed trunking switch pair.

Thanks Matthew_Fern.