There is an interesting thread on the MSDN forums about WWF Scalability. The question was how many concurrent workflows can a single workflow instance handle. The answer was the number of .net threads available to the process the workflow runtime is hosted in. Which appearently is around 25 for a single proc box and 100 for dual. According to Srikanth there is even a value on the constructor of the DefaultWorkflowScheduler object, which recommended value for which is 20 on a single proc box.
Looks like I get to give my WWF presentation again to CINNUG. The talk will be given at this month’s meeting on Tuesday the 21st. I have a few updates from Beta 2 which I didn’t realize when I gave the talk the first time (but have since been set straight, thanks James Conard!), and I will have one or two additional slides that I just didn’t have time for in the seventy minute time window at the code camp.
With the release of Beta 2 of WWF Microsoft has also provided us some new example labs. The labs for Beta 1.2 I thought were pretty good, so if you’re just picking up WWF these are a good place to start.
Also, Paul Andrew (part of the WWF team out in Redmond) has noted that he will performing a webcast in early Feb. about WWF. For those of you who had looked at WWF in Beta 1.