Scenarios and Best Practices › Scalability Best Practices › Scalability Recommendations › vCenter AIM Monitoring Recommendations › Scalability Limitations in Terms of Monitored Objects

Scalability Limitations in Terms of Monitored Objects

In general, CPU usage is the primary concern when scalability limitations are considered. For vCenter AIM monitoring, consider the three main factors that influence the CPU usage:

• The dynamic nature of the vCenter Servers being monitored.

  The level of activity of vCenter Servers influences the CPU consumption. The following scalability recommendations assume an average level of activity of the vCenter Server being monitored.

• The number of SNMP Managers, and polling intervals of those SNMP Managers.

  Large numbers of SNMP Managers polling the vCenter AIM, or short poll intervals, result in increased CPU consumption.

• The ratio of object count in relation to VM count.

  An object is any element of vSphere that is monitored by the vCenter AIM. For example, vCenter AIM monitors Datastores, Virtual Disks, Physical Network Interface Controllers, Virtual Switches, SCSI Controllers, ESX Host Hardware Sensors, and so on. The number of objects directly impacts CPU consumption. Due to the need to maintain the vCenter AIM cache and the additional overhead required for publishing this data. In real-world systems, there are typically from 6 through 11 times as many objects as there are Virtual Machines within a given vCenter.

Based on the preceding factors, a single SNMP Manager monitoring vCenter AIM with a 10-minute poll interval has a limitation of approximately 20000 VMs.