An instance is each individual installation of Meridian or Horizon. While there is no built-in limit on either, the answer comes down to individual network environments versus a simple count of nodes and alarms.
Scalability in an enterprise monitoring system rests on the amount of work required to monitor the inventory, which is measured in a more granular way.
Here are just a few things to consider:
- Port Monitoring: Yes/No? If yes, how many ports to check?
- Availability monitoring: Yes/No? If yes, how many IP services to monitor with synthetic transactions (ICMP, HTTP, DNS, etc.)?
- Performance monitoring: Yes/No? If yes, how many metrics per device in the inventory to collect?
- Threshold monitoring: Yes/No? If yes, how many IP services and performance metrics to threshold?
- Flow monitoring: Yes/No? If yes, how many flows per second per device?
Also factor in the frequency of each check and metric collection as well as the desired data retention period. The hardware or VM where you install OpenNMS and the various messaging and persisting dependencies impact the scalability of the deployment as well.
