So are you having a problem with the NPM server's ram/CPU maxing out, or with the routers and switches ram/CPU maxing out?
I guess a little more detail would be helpful.
If it's the server, the only advice I can offer is make sure you don't have SQL database on the same machine as the NPM software. I have similar resources on my server and my db has 15,000 objects in it - plus I have NTA on there too, which is a pretty resource-intense module. But I had to offload the database because it was bogging the machine down. SW recommends this too.
As far as the routers and switches are concerned, 53% memory usage isn't really a big deal. Did you make any changes to the switch/router configs themselves? Like maybe turn on netflow, QoS or SLA? If Orion is the only difference, what's your polling rate at and how many OIDs did you turn on - I have standard 2 minute polling intervals and the memory/CPU usage doesn't change much by turning on the Orion Services.
Thank you Quantus for responding
In fact I have two problems:
1 - the first one is the rise of the burden of RAM on routers and switches when I start NPM services
2 - the second one is the rise of the CPU load on the NPM server when I start NPM services
The following modules are installed in my server:
- NPM (SLX v9.5.1)
- APM (AL100 v3)
- IPAM (IPX)
- NTA (SLX v3.6)
- IP SLA (for 5 devices)
- NCM (DL200 v5)
Note: NCM is installed on a separate server than NPM.
The SQL Server 2005 (Enterprise) database is installed on a separate server too.
NetFlow is enabled on six routers.
- Nodes = 10min
- Interfaces networks = 9min
- Volume = 15min
When I start the Orion services, RAM and CPU climb significantly and my customer complains that the snmp queries saturate the resources of his routers and switches.