Increase in popularity of Citrix NetScaler brings a noticeable change in the sense that monitoring of the product becomes more and more important. This is where Citrix SCOM Management Pack for NetScaler (NetScaler Management Pack) comes into the picture.
NetScaler Management Pack is an availability and performance management solution that extends end-to-end service monitoring capabilities of Microsoft System Center Operations Manager (SCOM) to include Citrix NetScaler infrastructure.
A lot of questions regarding NetScaler Management Pack arose recently, specifically about its scalability, resource consumption, and performance. For this reason, our SCOM Management Packs team performed a series of activities, including measurements and analysis, to provide answers about the monitoring bandwidth of the product and its potential impact on monitored systems.
Our test-lab environment was as follows:
- One SCOM management server with one SCOM resource pool
- Two NetScaler VPX appliances
Since SCOM management server is used for NetScaler device monitoring, SCOM plays an important role in the entire process. It is therefore crucial to configure the SCOM infrastructure in the best possible way. For easier infrastructure planning, we utilized System Center 2012 Operations Manager Sizing Helper Tool. You can find this tool on the following webpage: https://blogs.technet.microsoft.com/momteam/2012/04/01/operations-manager-2012-sizing-helper-tool/
For actual measurements, Windows Performance Monitor was used. Two services were determined as crucial from the NetScaler Management Pack point of view:
- NetScaler Management Pack Agent (MPNSMonitorSvc)
- Operations Manager Agent/Microsoft Monitoring Agent (HealthService)
Our conclusions, based on the measurements and detailed analysis, are in short:
- The monitored objects count does not significantly influence the percentage of processor time used by the two services (MPNSMonitorSvc, HealthService).
- On the SCOM management server computer, there should be approximately 3 GB of physical memory available for the joint needs of both services (MPNSMonitorSvc, HealthService).
In situations with more than 14,000 monitored objects, we noticed substantial load on the HealthService service during workflow executions. The reason for this was large amount of workflows on HealthService which could not be processed and consequently caused data to be dropped.
Handle Count of HealthService (Operations Manager Agent/Microsoft Monitoring Agent) through time, measured in seven different validation sets
- Citrix SCOM Management Pack for NetScaler Performance Overview document, available here.
- NetScaler Management Pack product documentation is available here.
- NetScaler Management Pack installation package is available here.