Connecting Meraki Switch Stacks together for network redundancy

[michael-grosser]

I have a network that is a single point of failure nightmare, due to insufficient fiber installed. 

Stack A is a three stack of MS390-48UX connected to a Cisco 950048Y4C, lets call this DS01

Stack B is a three stack of MS390-48UX connected to a Cisco 950048Y4C, lets call this DS02

There isn't any spare fiber to run redundant connections from Stack A and B back to the DS01 and DS02, but Stack A and Stack B are in the same room, same rack in fact. 

The obvious solution is to connect stack A to Stack B so in the event of DS01 or DS02 is down, say for an IOS upgrade, the redundant link would prevent an outage. 

This configuration existed before, two stacks of Cisco 3850 switches connected back to Cisco 4500X distribution switches, but the link was removed when they converted to Meraki. I'm puzzled why, Shouldn't I be able to connection Stack A to Stack B, setting one side to Root Guard on the Meraki switch and Loop guard on the other side?  

[MHM Cisco World]

What you meaning link was remove?

MHM

[michael-grosser]

Stack A was a stack of three Cisco 3850 switch connect back to a 4500X DS01  

Stack B was a stack of three Cisco 3850 switch connect back to a 4500X DS02

There was a multimode fiber connection that connected Stack A to Stack B, this provided a redundant link between the two stacks in the event the connection between DS01 or DS02 was lost. This connection was removed and no longer exists. I want to restore this connection. It shouldn't matter what port on either stack i used, but for Maximum redundancy, it should be switch 3, fiber port 2 on stack A connected to switch 3, port 2 on stack B. 

[MHM Cisco World · Accepted solution]

There is no Issue As I see

If you have reach limit uplink to Core SW you can interconnect access SW for redundacy only.

What ypu must sure is this link is BLK by STP to make uplink to core SW always FWD.

For loop guard I will check and inform you if you can use in this case or not.

MHM

[michael-grosser]

The only think I would add to this discussion is the vlans are different for each switch stack, so the vlans will need to be spanned down to both switches from the distribution switches in order to have redundancy, otherwise the vlans will get blocked on the redundant link.