Wired Intelligent Edge

Hi community,

When deploying a small network, I usually place two 3810M switches with redundant power supplies. If there is a power outage at the site, I have redundancy, and if a switch fails or crashes (not the power), I still have redundancy because this switch goes down, but the other one keeps forwarding traffic.

Now I am going to place a 5406R switch for a new core, with redundant power supplies for sure. But in other to have redundancy in case a switch fails (not the power), is it enough to have one switch with two redundant modules? Or is better to have two switches stacked with VSF?

Thanks in advance,

Julián

Hi Julian,

I understand you are going to deploy 5406z and want redundancy.

The need of the "Redundancy" is defined well with below points:

• How important that network segment is to the network
• Number of user/Access Layer devices that might get affected
• How strong the technical team is presented in the location to recover the network incase of outage
• The End Customers budget concern
• There are other few questions that you might ask related to the case

In 5406R, you can have Redundant Management module (2nd J9827A) to have the module failure redundancy.(If budgetary becomes a concern over other questions above)

If you are looking for chassis redundancy, (assuming the aforesaid question have answers as Yes/Critical as answers) we can go ahead and propose 2unitsx 5406R

As of last time I checked, the Standby depoyment should be replica of the Commander Switch.

Hope the VSF document Link and the community link might help you. 

---

@Jeeva_selvakumar wrote: As of last time I checked, the Standby depoyment should be replica of the Commander Switch.

---

Not necessarily (it's not mandatory): both VSF Members can be differently populated with various interface modules (VSF Members, potentially, could be populated simmetrically or assimmetrically with MMs: say you can equip dual MM on 1st VSF Member and single MM on 2nd VSF Member or the contrary...an interesting discussion - still not fully answered - about having dual MM on VSF is available here)...there isn't a rule about keeping them exactly simmetrical...it's a matter of keep thing easy to manage.

Provided that all other VSF related restrictions and requirements are satisfied (a good read here) - at present time - the only other two suggestions I would follow are (a) to keep VSF Members as simmetrical as you can (for an easy management) and (b) to temporarily use just one MM per Chassis/VSF Member (see outcome of discussion referenced above).

@Jeeva,

This is going to be a core switch, so I think this answers all the questions you put, it is critical and there is no budget concern. The question is, if I have one switch with two management modules, am I fully protected against switch failures (regardless of power)? Because when a switch fails the likely root cause is the control plane, then having two management modules overcomes control/management planes issues...

@parnassus, what do you think about this regardless of the VSF symmetrically or asymmetrically deployment ?

Regards,

Julián

---

@fjulianom wrote: The question is, if I have one switch with two management modules, am I fully protected against switch failures (regardless of power)?

If you deploy one Aruba 5400R zl2 with dual MMs and you enable NonStop Switching redundancy model AND you have redunand PS Units that chassis is protected against one MM failure and/or against PS Unit(s) failure(s).

---

@fjulianom wrote: Because when a switch fails the likely root cause is the control plane, then having two management modules overcomes control/management planes issues...

Yes, generally if the Active MM fails and you have dual MM set in redudancy mode (Hot-Standby mode or NonStop Switching mode) the Standby MM kicks in and save your day...but an Aruba 5400R zl2 VSF deployment acts differently and MM redudancy is disabled (VSF and MM redundancy can't live together) once the VSF is formed...the additional MM you have on one or both the VSF Members is set inactive and it kicks in if the MM of that particular node dies (forcing that particular VSF node to reboot).