WCF ChannelFactory caching

Question

I've just read this great article on WCF ChannelFactory caching by Wenlong Dong.

My question is simply how can you actually prove that the ChannelFactory is in fact being cached between calls? I've followed the rules regarding the ClientBase’s constructors. We are using the following overloaded constructor on our object that inherits from ClientBase:

ClientBase(string endpointConfigurationName, EndpointAddress remoteAddress);

In the article mentioned above it is stated that:

For these constructors, all arguments (including default ones) are in the following list:

· InstanceContext callbackInstance

· string endpointConfigurationName

· EndpointAddress remoteAddress

As long as these three arguments are the same when ClientBase is constructed, we can safely assume that the same ChannelFactory can be used. Fortunately, String and EndpointAddress types are immutable, i.e., we can make simple comparison to determine whether two arguments are the same. For InstanceContext, we can use Object reference comparison. The type EndpointTrait is thus used as the key of the MRU cache.

To test the ChannelFactory cache theory we are checking the Hashcode in the ClientBase constructor e.g. var testHash = RuntimeHelpers.GetHashCode(base.ChannelFactory);

The hash value is different between calls which makes us think that the ChannelFactory isn't actually cached.

Any thoughts?

Regards

Myles

Answer 1

I'm on same boat and have the same question. Did you ever find an answer?

Answer 2

I too have had an issue with this, I get very quick performance when I save the proxy object for multiple calls.

What I really would like is to be able to create and use the proxy each call, but have the caching and optimizations occur behind the scenes.

Like you, I have followed the guidelines suggested by microsoft, including moving my binding configuration out of code and into a .config file (which I did not want to do).

I think this is something that should be handled by Microsoft in the architecture, it feels too much like I'm trading code quality for performance. If anything, they should provide us with a constructor which allows caching without having a presence in .config...