HPE Community read-only access December 15, 2018
This is a maintenance upgrade. You will be able to read articles and posts, but not post or reply.
Hours:
Dec 15, 4:00 am to 10:00 am UTC
Dec 14, 10:00 pm CST to Dec 15, 4:00 am CST
Dec 14, 8:00 pm PST to Dec 15, 2:00 am PST
Comware Based
cancel
Showing results for 
Search instead for 
Did you mean: 

IRF Stack question

 
eXPlosion_
Advisor

IRF Stack question

Hi,
I have an IRF stack of 2 HP switches:

A5120-48G EI JE069A

 

My IRF configuration:

irf-port 1/1
port group interface Ten-GigabitEthernet1/1/1 mode normal
port group interface Ten-GigabitEthernet1/1/2 mode normal
#
irf-port 2/2
port group interface Ten-GigabitEthernet2/1/1 mode normal
port group interface Ten-GigabitEthernet2/1/2 mode normal

 

Will I notice a difference if I remake configuration like this:

irf-port 1/1
port group interface Ten-GigabitEthernet1/1/1 mode normal

irf-port 1/2
port group interface Ten-GigabitEthernet1/1/2 mode normal
#
irf-port 2/1
port group interface Ten-GigabitEthernet2/1/2 mode normal

irf-port 2/2
port group interface Ten-GigabitEthernet2/1/1 mode normal

What is the difference between two? Thanks!

3 REPLIES
HP-Browniee
Respected Contributor

Re: IRF Stack question

Hi

The difference bewteen those  2 configs is:

With the first setup, you use a 20Gig conection between the 2 members because u bundled the 2 10G links in 1 irf-port. And if 1 link fails You still run on 10Gig.

The second setup you made is a ring topology. Your stack has  2 seperated links of 10Gig to the other switch. So you will never reach 20Gig through 1 link. But has the same redundency as setup 1, if one fails the other link is still active.

So the only difference is the amount of bandwith that passes trough the irf links. setup1: 20G setup2: 10G per link

I prefer the first setup, it has the same redundency  but with a higher bandwith.

Note: This only works with a stack of 2 switches.

 

I hope this made sense :)

Kind regards

parnassus
Honored Contributor

Re: IRF Stack question

Interesting question and answer.

Just three clarifications (hope to have not misunderstood the OP scenario!) about second case:

  • In a two members only IRF depolyment is it correct to define more than one IRF Port on each IRF member (no matter if that IRF port has one or more physical interfaces binded to it) to create redundancy at IRF Ports level?
  • The OP created Two IRF Ports per each IRF Member...that's strange if the IRF deployment has two members only. Is that right? It looks like a loop to me.
  • Partially related to the question above, formally a two members only IRF can't be deployed using an "IRF Ring Topology", so if the second way of configuring IRF Ports made by the OP (using Two separated IRF Ports per IRF member, each IRF Port binded to just one 10G physical interface instead of simply using One IRF Port per IRF member,  IRF Port binded to two aggregated 10G physical interfaces) resembles a Ring Topology...from what point of view is a Ring Topology?

So, IRF physical interfaces load balancing mechanism apart, in:

  • 1st case: we have, correctly, on each IRF Member a single IRF Port binded to two aggregated 10G physical interfaces <- Redundancy (if a physical link fails) and Load Sharing (enabled by default).
  • 2nd case: what we have really?

I too prefer the 1st case (it's the best practice!).

bakerbas
Advisor

Re: IRF Stack question

Best for a 2 switch IRF fabric is to use 1 IRF port on each device – put multiple ports in that IRF port for redundancy – this is a chain design and is best for 2 switch fabric

Yes – the OP used a loop as example for 2 switch IRF fabric – it works, but better to use chain for 2 switch IRF fabric.

No real advantage to using loop in 2 switch IRF fabric.