Finally performed an upgrade of three production based ipMonitor v8.5 systems. One system runs about 4k+ monitors, mostly SNMP and TCP/UDP port based. The second system runs about 2k+ monitors, mostly HTTP/HTTPS based.
Both always ran very low in regards to physical system resources. After performing the upgrade, we did notice a somewhat significant bump in the HTTP/HTTPS based system. Memory usage stayed about the same but the CPU processor more than doubled, granted this was about 8-10% before and now about 20-30%. To simplify our Alert and Maintenance schedules, we introduced a number of SmargGroups.
A total of 5 SmartGroups were created, one very large, almost 25 regular expressions to ADD monitors based on Display Name. Today, we had to delete this SmartGroup because logging into the interface was very slow and navigating anywhere was horrible as well. Since this was the only change to our system in regards to how we structure Alerts and Maintenance schedules, the "notification group" SmartGroup was deleted. This single SmargGroup was only assigned to one alert profile but figured that when a single monitor or even multiples would fail, since no hard relationship existed, it was taking a good amount of processing power to determine what alert to send as well as bogging down the interface.
Even though this fixed our problem, for now, we've obviously held off on using SmartGroups for Alert profiles. Has anyone else experienced this issue? And can any Solarwind developers comment on this behavior? It's very understandable that it would require additional processing to function this way, which is how we determined the cause so quickly.
PS. Single device, all monitors are assigned to this device as well. Multiple folders exist to structure dependencies only. Monitors are assigned to folders as needed, based on function/application.
Thanks.