Ready to buy Orion NPM - Reccommendations for larger installs

Question

We have 70 properties each with about 350 devices for a total of 2500 Wireless AP's, 500 switches and 300 PC's and 200 IMAC's.  Total number of devices is 3,000 to 3500.

After many demo's and setting up Orion for two sites we are preparing to recommend Solar Wind Orion SLX to our management and might add  a  APM and NCM to the mix.  We are sizing two pollers and hot stand by.

I would like to hear from folks that have similar size implementations and get some comments either positive or negative.

Thanks

Pacific · Answer

Thank you Dave, I appreciate the response.

savell · Answer

Hi Pacific,

We successfully run Orion, APM, Netflow, VoIP and a Hotstandby server with approx 4500 nodes and 16,000 elements (only 280 AOM component monitors at this stage).

We have two additional pollers (as well as the primary SLX poller). The poller hardware is fairly standard HP DL380 servers (actually G4 generation servers that are due for replacement).

The (separate) database server is a more modern 8 core DL380 server with 16Gig or RAM and a RAID card with 512Meg of RAM. The database runs on a RAID10 mirror across all available drives (8). It runs a 64bit version SQL Server 2005.

We also run the additional web server on the database server - we find this helps the web response times.

There are a couple of issues we face currently, but not show stoppers. One is that the pollers seem to have a low tolerance for losing contact to the database server. An outage of 30 seconds or so sees the pollers fail to recover and needing a restart (large memory growth in the netperfmonservice process). We will see what happens when our poller hardware is upgraded. It would also be nice to have better control over the poller load balancing process without having to restart pollers - given they take 15 minutes to restart, this is not something we want to do often.

Upgrades over multiple pollers can be painful (services fail to stop and then restart when you don't want them too) - but that is something I have got accustomed to dealing with.

The Hotstandby server is also limited at the current time (i.e. no module support), although I understand that this is something Solarwinds is looking at.

The other issue is Netflow performance. We currently only use Netflow for adhoc data capture as our disk I/O on the database server has difficulties in dealing with the data write activity. This is an issue with our database server specs, not an issue with Orion. We need to decide just how serious we are with netflow capture.

We also run NCM on another server, however only with about 2500 of our nodes at the current time. It's another solid product in my opinion, although I'm looking forward to some web interface and MAC/IP/Switch port inventory enhancements.

We have been very happy Orion customers for quite some time. The product continually evolves and improves. Coming from an Openview type environment, I can safely say that something drastic would have to occur to make us even look at going back.

Dave.