Wired Intelligent Edge

Hi all,

I’ll be having 3 5400 switches in a VSF stack. All sharing use of 5 VLANs for end users. Appreciate they’re all in a VSF stack so reduces the need for spanning tree - but is it good practice to enable it anyway?

Would you still have BPDU guard and portfast enabled on each port?

Anything else you would enable on each port as standard?

Thanks

---

@redford1980 wrote: I’ll be having 3 5400 switches in a VSF stack.

---

It's not possible: VSF requires and admits EXACTLY two (2) and ONLY two (2) Aruba 5400R zl2 switch.

 

Once VSF is deployed you can apply any known Spanning Tree best practice that fit your network topology best.

Ah ok - thanks for letting me know that.

So maybe 2 of them in VSF and then the third trunking all vlans over a layer 2 trunk link?

Thanks

---

@redford1980 wrote:
Ah ok - thanks for letting me know that.

So maybe 2 of them in VSF and then the third trunking all vlans over a layer 2 trunk link?

Thanks

---

Yes

 

and yes, you can continue to enable spanning tree for other stuff...

Well, it's a possible scenario: keep in mind that one major benefit of using VSF is that you add resiliency to network topology which is including it IF you configure uplinks to it from access (or others) layer specifically using LACP port trunks (aggregated ports) terminating to - distributed into - *both* VSF members...so if, for any reason, VSF splits (and it has a MAD mechanism in place to manage Split Brain scenarios) the connectivity to/from all layers interconnected that way is very well preserved (minimum disruption).

Thanks, I've attached option 1 & 2 ... would you say it's possible to have 1 LAG connection from both 5400 switches in VSF to another 5412 switch (option 2)?

 

Or would it be better to have 2 indivudal LAG interfaces to the other 5212 switch (option 1)? My only concern here is spanning tree ...

 

Thanks again

Option 2 is the correct one since VSF for your third Aruba 5400R zl2 recognizes the VSFs as just as single logical switch to connect to...so your Option 1 (Two LAGs) will form a loop and one LAG of two LAGs will be locked by (R|M)STP on VSF.

Within Option 2 diversificate, if possible, originating and terminating interfaces so you're protected also by modules faults on both ends.

Great - just to clarify; when you say diversaecate links in option 2 ... you mean spread the physical links between multiple line cards in case of a line card failure?

Thanks

Yes.

Say you have a 4 members LAG involving A and B modules on VSF ( of both members)...on VSF side:

1/A1+1/B1+2/A1+2/B1 (so it spans into A and B on both VSF 1 and VSF 2)

On 3rd 5400 say you have 4 modules involved, use:

A1+B1+C1+D1

Clearly it all depends on modules availability and interfacestypes (doing so using 10G interfaces could be very expensive - you need a lot of moduled - and having a 4x10G LAG could be overkill in some scenarios). Mine is just an (extreme) example.

Perfect, great reply thanks