I have written MbUnit tests which consume a SOAP service via a client generated by "Add service reference" in VS 2010.
The tests are run in parallel by using the DegreeOfParallelism attribute, and therefore have up to 10 tests running at once, each of which create there own instance of the Soap Client.
I have seen some really slow response time (c. 30 seconds) when running in parallel which do not occur when running a single thread. Before I start "blaming" the SOAP service, could I be facing problems because I am creating lots of clients in different threads?
Am I not using "Using", should I be utilising this? Would that make any difference?
Thanks
Robbie