Scalability recommendations

If your organization is expanding and looking to introduce additional elements and monitors, follow this guide on scalability recommendations. We've outlined the maximum counts and limits for resources and metrics in OpManager to maintain optimal performance. Use the OpManager System Requirements calculator and Disk Space calculator to get an accurate estimate of your system's required specifications.

Listed below are the maximum counts and limits for various elements and monitors in OpManager.

Interface count

We recommend monitoring up to 10,000 interfaces in a single installation for the Standard/Professional Edition. For the Enterprise Editions from version 12.8.456, up to 15,000 interfaces can be monitored per probe.

If the count exceeds the recommended limit, it will be efficient to increase the monitoring interval of those interfaces. Adding more interfaces will directly impact the overall performance of the product.

Note:

Interfaces with no data collection in the last 30 days will automatically become unmanaged. You can find them under the Data Collection Manager tab, located in Support → Diagnostics → Failed Interfaces.
You can avoid the addition of unnecessary interfaces by choosing appropriate criteria and conditions in the interface Discovery page.

Agent count

Without carrying out load-related configuration changes or affecting the network performance, you can install and monitor up to 1,000 agent-based devices for hassle-free network monitoring.

Group limit

OpManager restricts the number of elements per group to 2,000 to avoid any lapse in the performance of the product. Further elements cannot be added to the group if the limit is reached.

Business view limit

We recommend adding fewer than 500 elements in a business view for better visualization of devices in the MapView page.

Access point limit

OpManager can monitor a maximum of 2,500 access points at once. Regardless of the number of wireless LAN controllers in your inventory, the maximum cumulative number of associated access points that can be monitored is 2,500. In the Enterprise Edition, you can monitor a maximum of 2,500 access points per probe.

VLAN count

To avoid any hindrance in the performance of the product, OpManager limits the count of VLANs discovered to a maximum of 3,000. New VLANs will not be allowed to be discovered in OpManager past the specified limit.

Trap processing limit

To avoid any performance degradation in OpManager, the number of traps to be processed per hour is limited to a maximum of 50,000. If this threshold is breached, OpManager stops processing traps for a temporary period.

Recommendations for availability and performance monitors

Based on the monitor type/protocol being used with the performance or availability monitor, the maximum advisable number of monitors for a single installation are given below.

For versions below 128398:

Given below is the maximum advisable number of monitors for a single installation, when monitoring 1,000 devices.

Protocol	Monitoring interval	Polls/min
Availability	5	200
SNMP & Interface	15	1000
WMI & Application Monitors	15	300
CLI	15	150
VMware	10	1000
Xen	10	500
Maximum load (per installation)	10	1500

For versions 128398 and above (Enterprise Edition only):

Given below is the maximum advisable number of monitors for a single probe, when monitoring up to 2,500 devices.

Protocol	Monitoring interval	Polls/min
Availability	5	500*
SNMP & Interface	15	2000
WMI & Application Monitors*	15	1000*
CLI	15	250
VMware	10	1000
Xen	10	500
Maximum load (per installation)	10	2500

*Availability: A rate of 500 polls per minute is only recommended when using the "Windows Native API based Availability Monitoring(ICMP dll)" protocol for availability monitoring. For other protocols, a rate of 300 polls per minute is recommended.
*WMI & Application Monitors: A polling rate of up to 400 polls per minute is supported for DCOM, and 600 polls per minute for WinRM, subject to a combined limit of 1000 polls per minute. These limits apply to version 129135 and later. For earlier versions, a total of 400 polls per minute is recommended.

Monitoring best practices

Adding more monitors than the numbers suggested below will directly impact the performance of OpManager.
If it is required to add more monitors, then the polling interval of that monitor must be increased accordingly in order to balance the load on the OpManager server.
Adding monitors with more than 100 instances will indirectly affect the performance of OpManager.
Follow the recommended polling intervals below for optimal performance:
- Availability monitors: 5 minutes
- Performance monitors: 10 minutes
- Interface performance monitoring: 15 minutes

Recommendations for scaling up to 2,500 devices per probe in Enterprise Edition (from version 128398 and above)

For network devices: for a single probe, up to 2,500 devices can be monitored with agentless monitoring.
For server devices: it is mandatory to manage 1,500 devices using agentless monitoring and 1,000 devices using agent-based monitoring per probe.
Use only the default device monitors and the monitoring intervals provided in the Device Templates for optimal performance.
Availability polling must be done via the "Windows Native API based Availability Monitoring (ICMP dll)" protocol.
Ensure that network latency is less than 2 seconds to maintain smooth operations.

Syslog processing limit

A syslog rate of 200 syslogs per 2 minutes is recommended to avoid any hindrance in the product performance.

NetFlow processing limit

If the NetFlow processing limit exceeds 1,000 flows per second, it will affect OpManager's performance. Ensure the NetFlow processing limit is within the recommended rate.

Firewall logs processing limit

Ensure the Firewall log processing limit is under 1,000 logs per second to avoid any lapse in the performance of the product.

OpManager Enterprise Edition

Monitoring bandwidth

OpManagerEnterprise Edition monitors multiple remote sites using OpManager Central and probe servers. The maximum number of devices monitored by OpManager Enterprise Edition, maximum number of probes supported, and related limits are listed below.

Maximum number of devices monitored	10,000
Maximum number of interfaces monitored	50,000

For Enterprise Edition with Lite Central enabled

OpManager Enterprise Edition's Lite Central option helps you retain the archived data at probe level, helping you optimize performance by reducing the load on the Central server's database.

Maximum number of devices monitored	30,000
Maximum number of interfaces monitored	100,000

Number of supported probes

For OpManager Enterprise Edition's Lite Central option, a maximum of 100 probes will be supported.

Maximum allowable latency

Ensure that the latency between OpManager's Central server and a connected probe is within 50 ms (both ways). The Central server can still function at a higher latency, but when the latency crosses 100 ms, its performance can be affected.

Maximum number of alerts processed

The probes send alarm details to the Central server, which is capable of processing 1,000 alerts per minute at peak performance. Note that the Central server is capable of handling more than 1,000 alerts per minute, but this may have a slight impact on its performance.

If you need to monitor more than the mentioned scalability recommendations, please contact our support team at opmanager-support@manageengine.com.