We've set up our Zabbix 1.8.4 server connected to a dedicated Oracle server running release 11g. This release includes database resident connection pooling, which we've turned on in hopes of decreasing database server load and increasing performance.
Thus far it works ok and seems to reduce database load. All I did to enable it was to add the ":POOLED" statement to the service descriptor (compiled it in). Our DBA did the settings on the database side.
Has anyone else tried this? Experiences pro or con?
Also, we seem to be able to collect up to about 450 items/second with this setup, and we have plenty of hardware overhead left. Does anyone have a suggested configuration for this many hosts?
Our current stats:
Number of hosts (monitored/not monitored/templates) 1286 1205 / 19 / 62
Number of items (monitored/disabled/not supported) 67324 40500 / 23542 / 3282
Number of triggers (enabled/disabled)[problem/unknown/ok] 18671 16578 / 2093 [106 / 404 / 16068]
Number of users (online) 31 3
Required server performance, new values per second 316.34 -
Erik
Thus far it works ok and seems to reduce database load. All I did to enable it was to add the ":POOLED" statement to the service descriptor (compiled it in). Our DBA did the settings on the database side.
Has anyone else tried this? Experiences pro or con?
Also, we seem to be able to collect up to about 450 items/second with this setup, and we have plenty of hardware overhead left. Does anyone have a suggested configuration for this many hosts?
Our current stats:
Number of hosts (monitored/not monitored/templates) 1286 1205 / 19 / 62
Number of items (monitored/disabled/not supported) 67324 40500 / 23542 / 3282
Number of triggers (enabled/disabled)[problem/unknown/ok] 18671 16578 / 2093 [106 / 404 / 16068]
Number of users (online) 31 3
Required server performance, new values per second 316.34 -
Erik
Comment