Application Dependencies requirements in SAM
Before enabling the Application Dependencies feature to track how applications, application processes, and servers connect to each another in your environment, review the following requirements and recommendations.
Each dependency must involve two separate physical or virtual servers, represented as nodes in the Orion Web Console.
Each node must be monitored by at least one assigned application monitor.
Do not enable this feature on the Orion server or a scalability engine such as an Additional Polling Engine.
Linux distributions must support GNU C Library (glibc) 2.18 or later.
The following environments are not supported:
Do not use IPv6 or link-local addresses to communicate within the network segment (link) or the broadcast domain to which the host is connected.
|Orion Platform agent||
Enable "Allow automatic agent updates" on the Manage Agents > Edit Agent Settings page so you don't have to update agents manually.
Remote systems that host applications and listen on a specific port must support the following exceptions.
Using this feature with Windows Failover Clusters (WFCs) has additional requirements.
For optimal performance, use the Application Dependencies feature to monitor up to 500 nodes. For example, in a typical deployment where 500 nodes are monitored, and another 500 nodes are not monitored by this feature, you can track dependencies for:
- Up to 7500 application-to-application connections, and
- Up to 1500 node-to-node connections.
Note the following details about these guidelines:
- Recommended limits are not cumulative. The number of application-to-application dependencies does not impact the number of node-to-node dependencies.
- These limits assume an average of 3 IP addresses (IPv6 or IPv4) and 5 applications per node.
- Orion Platform agents hosted on minimally provisioned servers can consume high CPU usage during polling. If that occurs, reduce the quantity of monitored component monitors or increase server resources. For more tips, see SAM polling recommendations.