OPC Client performance

This section describes the relationship between CPU usage and Scan Groups, versus Object usage.

Industrial Application Server (IAS) was configured with OPCClient Objects accessing a DAServer on a remote node. The DAServer was configured to retrieve changing data from a PLC via an Ethernet connection. The items with the OPC Client Scan Groups were accessed via Object Viewer.

Note The test results described in this section do not take into consideration implementation of Object Scripts, which will affect the execution of the AppEngine; nor do the results account for limitations in the number of items which can be accessed through a single object.

The performance test was conducted with the following basic topology:

The test was conducted using the following strategies:

1. Establish "Base Line" CPU consumption.

   1. An OPC Client Object with 2,500 items in 5 Scan Groups was measured.

   2. Items were then increased to 5,000.

   3. Items were then increased to 10,000.

2. Measure CPU consumption when increased scan groups are used.

   4. An OPC Client Object with 10,000 items in 5 Scan Groups was measured.

   5. The 5 Scan Groups were evenly split into 10 Scan Groups.

   6. The 10 Scan Groups were evenly split into 20 Scan Groups.

3. Measure CPU consumption when increased OPC Client Objects are used.

   7. 5 OPC Clients with varied number of items, totaling 10,000, were deployed.

   8. The 5 Objects were split into 10 objects with item total remaining at 10,000.

   9. The 10 Objects were split into 20 objects with item total remaining at 10,000.

After the Objects were deployed, the system was "stabilized" for 5 minutes to attain CPU usage. The following results are represented in percentage of total system utilization:

The use of multiple Scan Groups appears to have more impact on system performance than does multiple items, although under these test conditions, each are negligible.