Any documentation or experiences on configuring SNMP agent
in NCS to receive polling requests?
We are looking to use Cisco Prime Performance Manager for
device discovery and data collection from NCS for our VMS Service Assurance
Solved! Go to Solution.
Its really time to stop doing new development with SNMP. It has lived a useful life, but has many problems and should be allowed to die gracefully. We can make choices with new development to use other protocols that will save Cisco and our customers time and money in the long run.
Unfortunately, we are likely at the mercy of Cisco Prime Performance Manager capabilities - it could only support SNMP or SSH or Telnet as the device management protocol.
Perhaps, the latest release has added Netconf and HTTP/S. Durga, could you please double check ?
If so, the amount of work to use SNMP to read data from the NCS /devices/device tree is going to be substantial. MIBs have to be produced from the NCS device tree. A bit of work,
How to go about to do this is covered in the NCS UG, since no one ever tried to do this with DMs a complex as the NCS device tree, this work is likely to have a few bumps in the road, i.e the generated MIBs might be a bit rough.
The basic idea is that we can take any yang model, generate MIBs from them that has built in instrumentation to read from the yang/cdb database.
Basically - just a bad idea.
Thanks Ajay and everyone for input on this. I am able to get to the point where PPM has discovered the Tail-f NCS agent.
Rajiv - I will get back with info. I find out - I think PPM can manage using netconf but supported devices may be limited.
Keeping in mind PPM is more interested in the operational data and KPIs while NCS mainly interested in the configuration data. Not sure NCS is meant to feed/collect the depth of PPM KPI collections…
Yes -- but the weird place we are right now is that NCS is collecting perfdata from the VNF's. I am not saying its good or bad, just where we are right now. We have had several issues already with mixing up deployment and configuration with collecting perfdata. My $0.02 is that using NCS to collect perfdata via snmp wont scale. The portal displays perfdata reported by NCS that is collected by SNMP.
I would vote that perfdata for VNFs should be handle and come from PPM piece within the ESC app that’s anyhow should be the case for supporting the elasticity of the services
I agree, letting the PPM in ESC get all the perf data from the VNFs may be a reasonable approach.
However, given that the Service Assurance stack would need to get the perf data for analytics anyway, why not let the Service Assurance Stack get the perf data directly from the VNFs and pass theinfo/trigger to the ESC (to deal with elasticity), as/when needed. Why make ESC the bottleneck for all perf data ?
Holistically, the triggers for ESC moving the VMs may quickly be beyond a VNF being unreachable. So, it would be useful to have the northbound system provide the additional triggers.