Do devices connected to MS350 fail over network and poe to warm spare without moving cables?

[Nick_Todd]

Hi, I am considering purchasing 2 Meraki MS350-48-FP switches with the plan to use the warm spare feature. I will have around 30 devices in a small office connected to one switch. Around 20 connections will be poe cisco phones that use the pass through ethernet on the phone to connect laptops / pcs on user's desks. Other devices will be a few printers, MR wifi APs, Logitech rally bar and tap controller VC equipment.

I haven't yet read any info on whether i actually still need to move the network cables for the devices from the primary to the spare switch in event of a complete failure of the primary. Or are the ports on the failed primary still operational and powered by the warm spare?

If I don't need to move any network cables from the failed primary, does the warm spare take over powering any poe devices, ie the phones, that are connected to the ports on the failed primary?

Thanks,

Nick.

[DarrenOC · Accepted solution]

Morning @Nick_Todd, if the switch fails that your Endpoints are connected to then all those Endpoints will lose connectivity.

[DarrenOC]

Please read the below guide.

https://documentation.meraki.com/MS/Layer_3_Switching/MS_Warm_Spare_(VRRP)_Overview

[Nick_Todd]

Thanks, I had looked at that article. The 'spare' feature seems to be just configuration that would transfer to another switch. I wish they would invent some way to share the ports to other switches for physical failover of the  endpoints ! 

[GreenMan]

The only way to do this would be to have PCs etc with dual Ethernet NICs and attach to both switches simultaneously.   You might then be able to do NIC teaming, if the two switches are physically stacked and you set up a Link Aggregation Group on the stack.   Typically this is done solely for servers - no-one ever really does it for user access layer.  In the event that a switch fails,  you might re-patch some key devices (APs?) quickly to an adjacent switch but more realistically - replace it with a maintenance spare.

As Darren mentioned, the warm spare feature is not designed for this

[Nick_Todd]

Yes, I have servers set up with teaming currently across 2 switches. Current set up is 2 Cisco CBS 350 48 port switches with 4 ports LAG to link them both. All devices can fit on one switch if needed, but currently distributed across the 2 switches. Failure recovery is just to physically move the endpoint network cables to ports on the other switch. Its a pity that there is no real fail over option for switches, a second switch to somehow take over the traffic on a port on another switch if its board, or power failed.

[GreenMan]

I think it's just cost-prohibitive for most customers, given the cost of switches & how reliable switches are.  Also - most client devices simply don't have two NICs

[cmr]

@Nick_Todd if they are laptops then they will connect to the Wi-Fi if the wired Ethernet fails, just make sure the APs aren't all on one switch!

[GreenMan]

Just to say too - this isn't a Cisco or Meraki thing, it's industry-wide, AFAIK

[PhilipDAth]

Using the switches in stack mode is so much easier.  Any reason why you want to use warm spare mode?

>Yes, I have servers set up with teaming currently across 2 switches

Note that you can't use LACP channels across switches unless they are in a stack.  This doesn't work when using warm spare mode.

[Nick_Todd]

Hi Philip, no reason to use 'warm spare'. I have been on the lookout for a way to have some redundancy on the network patches that didn't involve physically moving cables, warm spare sounded like what i was looking for. My background is in other infrastructure, not so much with switches etc. As i understand, a stack of 2 switches means it operates as one switch. Does this mean the ports are controlled by either of the switches? Can one switch die and all the ports on both switches are still running and controlled by the other switch in the stack? What Im looking for is automatic failover in some manner. If a switch fails i want another switch (or something) to take over and continue running the network for the devices connected to the failed switch. Is that possible? Im trying to have no physical actions like cable moving, all automatic.

[cmr]

@Nick_Todd warm spare in Meraki switches is just to make the manual switch over quicker, with the L3 routing moving from the failed switch to the spare automatically.

I've never seen an automatic repatching system, though it could exist of course, but it would not be a switch feature, rather a patch panel feature with a physical rerouting of the active cable pairs.

[cmr]

I think this is the product you need, though you need one switch perhaps cable that you want to automatically fail over:

https://www.valiantcom.com/aps/safecomm/safecomm-gigabit-failover.html

[PhilipDAth]

>Hi Philip, no reason to use 'warm spare'

Use the simpler "stack" mode.  It has a lot of advantages.

>to have some redundancy on the network patches that didn't involve physically moving cables

The only way I know of to do this is to go to a chassis-based switch with line cards.  These have separate "supervisor" engines, and you can have a supervisor die and all the ports stay up and working.  No re-patching.

There is nothing in the Meraki lineup that does this.  The cheapest option would be a Cisco Catalyst 9504R.  You can have up times of 10 years on these kinds of switches.  They are "high availability" in every sense.  You can even upgrade the firmware without any outages (* when it works, and you need a licence that includes ISSU).
https://www.cisco.com/c/en/us/support/switches/catalyst-9400-series-switches/series.html 

These start at USD$50k for a minimal configuration, but it is pretty easy to spend USD$100k on one of these configurations.  Maintenance costs of 10% per year should be expected.

This is not a solution you should spec up yourself due to the complexity.  If interested, get a Cisco partner involved.

https://locatr.cloudapps.cisco.com/WWChannels/LOCATR/openBasicSearch.do 

>stack of 2 switches means it operates as one switch.

Correct.

>Does this mean the ports are controlled by either of the switches?

No.  The ports that the switches are contained in are purely controlled by that switch.  Switch management is performed by one of the switches, and will fail over to the other switch if the primary dies.

>Can one switch die and all the ports on both switches are still running and controlled by the other switch in the stack?

No.  If a switch dies then all of the ports in that switch will die.  The Cisco Catalyst 9504R with dual supervisors can do this though.

[Nick_Todd]

Thank you all for your information. Given the high costs, I think I will be sticking with the manual moving patch cables to the second switch.