We're running one NPM poller with a separate web server and separate SQL server. The SQL server is an older box (Dell PE2850) with 8GB RAM and local-attached disk. When we've been running WITH Netflow service enabled, the average disk queue length (average) has been getting up to around 45, and this server has 2 physical disks. Needless to say, we run into problems with monitoring.
I've been working heavily with our DBAs and learning a ton about SQL, and I already have a case open with Solarwinds Support (119612) to address the LDF file size growth (it's out of control when Netflow service is enabled on the poller).
Meanwhile, it's fairly obvious that our SQL server isn't beefy enough to handle the load. What do I need to ask for? I've heard Josh (headgeek) talk about Ramdisk as one way to boost performance, but the server guys look at me with a blank stare as if to say "why the hell would you want ramdisk?" Should we ask for solid state disks, even though they might be really expensive?
Thanks in advance for everyone's help on this! Any suggestions are welcomed!
Swack
NPM SLX License v9.5 SP4
node polling: 300 sec
interface polling: 600 sec
volume polling: 300 sec
rediscovery every 60 minutes
statistics collected every 10 minutes for nodes, 10 minutes for interfaces, and 15 minutes for volumes
30 days detailed statistics retention
60 days hourly retention
180 days daily retention
120 days events retention
1600 nodes
5151 interfaces
135 volumes
NTA v3.5 SP2
Would like to monitor at least a dozen routers
