There are several factors that contribute to the final performance, e.g.:
How fast can the underlying EasyXXClient component supply data.
Any additional penalty introduced by the "plumbing" in the Reactive Extensions for OPC, or StreamInsight adapters, payload classes etc.
The actual processing, calculations and decisions made on the data, be it in StreamInsight or in user's code.
As to #1, the same component is used as with any other model, so this point does not contribute to any difference in speed.
Regarding #2, the "plumbing code" is quite thin, and the expected amount of overhead it creates is minimal.
Item #3 we cannot influence as such, and it is the actual useful work being done, so it is questionable whether that should be considered into performance measurements of this specific approach.
This said, I have no numbers available that I can give to you currently. If you deem necessary, we can set up a test environments and perform basic measurements; it would take some time and effort though.
I am starting to review the details of the StreamInsight interface [...]. I have a quick question on this.
Have you done any performance testing of the amount of data that can be processed using the Reactive programming model and/or streaminsight versus the standard OPC subscriptions?
I am trying to get a clear understanding of the benefits of using these methods to create a client application versus the traditional approach.