[c-nsp] RSP failover vs Chassis failover for switch/router clusters

[George Giannousopoulos]

Hello all,

I'd like to hear the community's opinion and experience when working with
switch or router "clusters". By saying cluster I assume any kind of closely
coupled systems, active/active or active/standby, such as VSS, vPC, mLACP.

Also I assume that the cluster members will be physically close to each
other and all uplink/downlink devices will be dual homed on both cluster
members.

Given the above, I wonder if such systems should be equipped with dual RSPs
or not.

In case of dual RSPs, any RSP failure will trigger the standby RSP and the
device will try to maintain forwarding by transferring the execution of
processes to the previously standby RSP.

In case of single RSP, any RSP failure will trigger the clustering
mechanism and ideally the other member of the cluster should take over.

Would you prefer a chassis or an RSP failover having in mind the highest
possible service availability?

Thanks,
George