IRF Option

[GlenRB]

I recently set up an IRF on two 5130 switches and I'm confused about some options.  One sample config showed to put two interfaces into an irf port on each switch.  For example:

irf-port 1/1

port group interface ten-gigabit 1/0/51

port group interface ten-gigabit 1/0/52

 then on switch 2

irf-port 2/1

port group interface ten-gigabit 2/0/51

port group interface ten-gigabit 2/0/52

On another sample config it said make two irf ports for each switch.  For example:

irf-port 1/1

port group interface ten-gigabit 1/0/51

irf-port 1/2

port group interface ten-gigabit 1/0/52

and on switch 2

irf-port 1/1

port group interface ten-gigabit 2/0/51

irf-port 1/2

port group interface ten-gigabit 2/0/52

I did the first example which seems to be working but I'm wondering the real difference and pros and cons of each example and a little clarification. 

Thank you.

[parnassus]

Considering your scenario (only two IRF Stack members involved) you did it correctly IMHO.

Read this thread discussed few months ago: the OP's Title will initially look unrelated to the IRF configuration but the interesting discussion that followed was indeed focused about the "two members IRF Stack" proper configuration to adopt and it will shed some light about the differences you've found in setting up a two members IRF Stack.

Hope it helps.

[GlenRB]

Thank you for the information.

Would your recommendation change if this were a three member IRF stack?

[parnassus]

Hope I understood your question correctly: No HPE recommendations will not change.

Since you are wisely using the IRF Link Redundancy at logical IRF port level in your Daisy Chain IRF Topology...the point, if you are going to add a third Switch to your IRF Stack mantaining the link redudancy you deployed, is to migrate it into a Ring IRF Topology (someone will say that HPE supports Daisy Chain IRF Topology with three members but it also doesn't recommend it favouring the Ring Topology when members are three or more) maintaining the IRF Link Redudancy configuration you already correctly deployed (remember you have two physical ports binded for each logical IRF port defined on each switch)...this will require (1) to define a new logical IRF port on each switch and (2) to use another two physical ports binded to it on each switch...so you will need to grant a grand total of 4 physical ports (as example) dedicated to IRF links per switch (and that is the point: you need to have them available) and, finally, also (3) consider new cablings to accomodate the changed IRF topology with the insertion of the third IRF Stack member.

[Apachez-]

Here is a good summary on the two modes of IRF (daisy chain vs ring topology): https://www.dasher.com/a-deep-dive-into-hp-intelligent-resilient-framework-irf/

[parnassus]

One adapted statement (similar to one found into IRF Configuration Guide), when IRF is made of three or more members, to rule them all:

"Use a IRF Ring Topology and aggregate more physical ports to a single IRF Port for implementing IRF Link Redundancy, this will provide IRF link resiliency and increase bandwidth."