Catalyst 3860 Redundancy Options

[Jeffrey6413]

Hi All,

I wanted to check-in with everybody and make sure I am not doing this setup any harder than it needs to be. 

What I Got:

• One Cisco Catalyst 3850 Switch acting as the “Core” switch. Layer 3 for vlans sit on this device.
  • We have two Dark fiber Paths: North to Data Center, and South to another building which also has redundant paths to the data center.
  • OSPF runs the entire operations
  • VRF for guest-networking, keeps things separate, nice and clean.
• One or More Stacks of Cisco 3650 switches as the access layer. Switches use the Lan Base License.
  • Each stack has 2 x 10GB uplinks in a port-channel configuration back to the core. This is for redundancy mostly.
• Spare 3850 switch

Where we need to be:

• Need to add the second 3850 switch to help create redundancy. We want to be able to do things such as IOS updates without taking down the entire location. Supporting Multi-Chassis Ethernet channels would be great, but we can work without it.

Some problems I have:

• I considered using the traditional stackwise setup, but I have found on the 3650 setup that this tends to be very difficult. I have been unsuccessful in running switches within a stack on even slightly different IOS IE: 16.6.0 vs 16.6.1. This makes it hard when doing patching
• I considered the Stackwise virtual but looking over the documentation, both 3850’s would be a mirror of each other. The problem there is I do not have redundancy links north and south

My solution would be to do a simple HSRP on the vlans, and OSPF between the two switches and keep them separate. T Can put one dark fiber link on each unit. I would have to deploy another set of uplinks to the 3650’s, which would not be a problem. This would keep everything available, and allow me to take various sides of the core down without impacting everything. 

Am I missing something?

[paul driver]

Hello

---

@Jeffrey6413 wrote:

• I considered using the traditional stackwise setup, but I have found on the 3650 setup that this tends to be very difficult. I have been unsuccessful in running switches within a stack on even slightly different IOS IE: 16.6.0 vs 16.6.1. This makes it hard when doing patching
•  

My solution would be to do a simple HSRP on the vlans, and OSPF between the two switches and keep them separate. T Can put one dark fiber link on each unit. I would have to deploy another set of uplinks to the 3650’s, which would not be a problem. This would keep everything available, and allow me to take various sides of the core down without impacting everything. 

---

The first option above would be the most applicable to do and simplest solution, The second option is also seems viable but a lot more administrative - , Can you elaborate on what issues you've had or envisage running the 3860's in stack?

[Jeffrey6413]

Anytime a switch has come up in a 3650 StackWise stack with a slight different IOS in: 16.6.0 vs 16.6.2, it would give an error about version mismatch and not come back up in the stack.  So when I did IOS updates, which I try to do about every 3-6 months, I have to reboot the entire stack.