0.7 (WIP) Better Performance for Read/Write Operations

Batching Read and Write Requests for Better Performance - Work-in-Progress

Release 0.7 adds new low-level handling for single-node read and write requests. (Serving requests coming from periodically scanned or passive records.)

Instead of executing each request instantly with each request getting its separate OPC UA service call, the requests are queued (three prioritized queues controlled by the EPICS record PRIO field) and a separate worker thread works off these queues and creates batches of requests that are combined into a single service call.

Throttling parameters can be set through the opcuaSetOption interface. If both *-timeout-min and *-timeout-max are set, the wait time is interpolated depending on the actual number of nodes in the batched request.

Option	Setting
nodes-max	maximum number of nodes for any service call
read-nodes-max	maximum number of nodes for read requests
read-timeout-min	minimum time to wait (in ms) after a read request
read-timeout-max	time to wait (in ms) after a read request for the maximum number of nodes
write-nodes-max	maximum number of nodes for write requests
write-timeout-min	minimum time to wait (in ms) after a write request
write-timeout-max	time to wait (in ms) after a write request for the maximum number of nodes

Please also see the list of issues closed in v0.7.0 and the changes since v0.6.0.

As always: please report any findings in the issue tracker