08-27-2021 09:20 AM
New to the 9200X line of switches. I'm trying to figure out the best way to uplink a single 9200X to two seperate 4500X switch stacks. Links don't have to be active/active load balancing (though that's preferred). If approach is active/standby I'd need to be able to at least do some simple logic (ping IP) to determine which link to use.
To be clear, I have two pairs of 4500X switches, each in their own VSS stack. Those two stacks connect to one another across a multichassis etherchannel trunk. I need to connect a 9200X to both 4500X switch stacks in a manner where a single 4500X stack failure (or reload) won't cause an outage.
If this isn't possible, would using a stacked pair of 9200X switches make a difference?
Solved! Go to Solution.
08-30-2021 09:04 AM
Hello @lonelyadmin ,
if you build OSI L2 port channels you can simply rely on Spanning-Tree ( a form of Rapid STP Rapid PVST or MST) and you can have VSS1 root bridge and VSS2 the backup root bridge so STP will keep in blocking alternate the port-channel to VSS2 and in forwarding state the port-channel to VSS1.
The 9200 catalyst needs to run the same flavor of STP.
if you have L3 routed port-channels you can use a routing protocol like OSPF and you can even achieve load balancing and redundancy anot not only fault tolerance like in the L2 case using STP.
Hope to help
Giuseppe
08-27-2021 10:39 AM - edited 08-27-2021 10:40 AM
So, the issue with this design is that if you connect a 9200 to different VSS sets, you can't use any Portchannels as the 9200 will see each VSS set as a different device and not one logical device/domain. The same problem also persists if you have a stack of 9200. I think the easiest way for this to work is to connect the 9200 (single or a stack of 2) to one set of VSS since you already have redundancy between the VSS sets.
HTH
08-30-2021 08:22 AM
That's pretty much what I've been doing in other places with similar setups. When there is an outage I usually have to either physically move cables if the link to the other stack doesn't exist or shut/no shut ports.
I understand that real multichassis portchannels likely will never happen on these class of switches. I'm wondering if I can go ahead and physically cable up to both stacks, each with their own portchannel and then manage the links from EEM or kron. I haven't done any work on the 9200X and don't know if there are any licensing hurdles...or even what capabiilities it has in that regard. But I think I'll look at finding a way to monitor the uplink to check for direct connectivity (CDP on port-channel?) and some remote check (icmp through the parent switch?) or what ever logic I can come up with and have the interfaces shut/no shut appropriately. Having an external monitor shell in or SNMP control it is possible as well.
Having a short outage while links change over is ok. Better than waiting on me to identify and verify the problem and then move cables or whatever.
08-30-2021 09:04 AM
Hello @lonelyadmin ,
if you build OSI L2 port channels you can simply rely on Spanning-Tree ( a form of Rapid STP Rapid PVST or MST) and you can have VSS1 root bridge and VSS2 the backup root bridge so STP will keep in blocking alternate the port-channel to VSS2 and in forwarding state the port-channel to VSS1.
The 9200 catalyst needs to run the same flavor of STP.
if you have L3 routed port-channels you can use a routing protocol like OSPF and you can even achieve load balancing and redundancy anot not only fault tolerance like in the L2 case using STP.
Hope to help
Giuseppe
09-01-2021 02:09 PM
I like that idea...I sometime forget about those. I do have HSRP running on all of the L3 interfaces.
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide