I am wanting to perform a baseline of our core network performance. What i am particuarly interested in is how many packets per second the core switches are processing and the current backplane performance. We are using older 3550 stackable switches in our core. I am measuring CPU and interface statistics but I dont think this is providing me with the information i require. The CPU of the core switches sits at a very low 3% on average and looking at the interface statistics i cannot see any drops or failures.
Any advice would be appreciated.
You should be very happy and dancing for such a low-traffic network.
By the way, what tool do you use to do monitoring?
You may like to try MRTG [Freeware] or PRTG to monitor things like CPU/Interface utilization, Error rate etc.
I use cacti and PRTG funnily enough for monitoring. PRTG to monitor netflow information and cacti for the cpu / interface statistics.
The reason i ask these questions is because we are setting up a seperate redundant site and i am trying to toss up whether we should go for a stacked or modular based solution.
Establishing a baseline for network metrics will typically only come through manual observation. I have also worked on networks where the number of endpoints and flows are strictly defined meaning we could calculate the baseline before we switched the network on.
Once you have a baseline defined a good tool for monitoring is Opsview as it allows your to define alert thresholds for each collected metric, either globally or per-device SNMP OID, alerting once crossed.