Considerations For Building Highly Available SolarWinds Monitoring And Alerting Solution

Hi Alex,

I've deployed our NPM with platform resilience in mind.  My application server sits on a VMWare cluster that spans two data centers as does my SQL database.  Unfortunately I've not come up with a way to effectively have HA from an application perspective yet.  I have deployed a remote polling agent in our other European data centers that do the polling of remote nodes and report their information back to the core node.

When your doing your planning you need to bear in in mind that there is a hard limit of 10,000 nodes per NPM server that can be monitored.  That can be increased by deploying more polling agents.  And there aren't any limits (that I know of) around the total number of polling engines you can have.  So if you've got a SL2000 license you can have 2000 polling engines, although that would be a little over kill :-).

Mark

Have we perhaps looked at deploying FoE? http://www.solarwinds.com/extensions/failover.aspx

Good FoE overview: http://www.solarwinds.com/resources/videos/orion-failover-engine-overview.html

We use FoE for our primary and additional engines (the additional poller is for scale, not redundancy), and NeverFail for the DB server (same technology).  We keep one set of VMs in each data center.  Cool stuff, although it will add time to your patching and updating processes.