The run time of discovery could vary drastically depending on the type of discovery, the scope, the size of the environment, the server in which the discovery node component is installed, and the number of nodes assigned to the cluster.
To improve performance in the discovery process, it is recommended to:
- DO NOT install the discovery node component on the same server as the Enterprise Reporter server component or SQL Server host, as this will cause degraded performance.
- The Enterprise Reporter server and the node(s) communicate directly with the database frequently, so locating them close to the SQL Server and choosing a SQL Server version with enough power to manage the connections and data transfer from the servers is recommended for best results.
- When a discovery is run on a cluster, the Enterprise Reporter server assigns work to its nodes. You can add nodes to a cluster at any time. Each node can only belong to one cluster. You can increase the performance of your discoveries by ensuring that nodes are configured to optimize the maximum number of concurrent tasks or by adding new nodes.
- Assign the discovery to the cluster geographically closest to the targets. If necessary, break a discovery into smaller discoveries to accomplish this. By assigning each discovery to a cluster, you can also control which nodes are used for each discovery.
- Group together targets in a discovery based on the data you are collecting. A discovery should collect generally the same data from all targets. This makes it easier to design and maintain discoveries.
- It is recommended to set tasks to be System-managed to optimize node performance. CPU load is always used to determine how many concurrent tasks are assigned. You may optionally enter any number greater than 0 to set a Maximum number of tasks that the node can process concurrently.
- Verify the scope selected in the configuration of a collection task, a big scope or some specific settings may complicate the amount of time it takes the discovery to complete.
- Each collection task is assigned to a node, balancing the distribution across the nodes until all the nodes are processing as many tasks as they are able. If no nodes are available to process the task, the task must wait until a node becomes available. Node performance is based on a combination of memory, processor speed, disk space, and network bandwidth.
For more information, please review the Installation and Deployment Guide for Enterprise Reporter:
https://support.quest.com/enterprise-reporter/technical-documents