Saturday, January 29, 2011

WCF REST and .NET 4.0

REST  and SOAP have long been available as specifications how to access and interact with services published on the internet (or intra- /extra- net). Yahoo, Google and Amazon Web Services have long been using REST as the primary way to expose different APIs. In the .NET world we initially had SOAP based web service support built into Visual Studio (.NET / 2003 / 2005), meaning that we could point Visual Studio to a WSDL file and generate the needed code to interact with this service. You could use a REST based web service using the .NET framework by manually using a combination of HttpWebRequest, HttpWebResponse and generating supporting classes for the returned XML using xsd.exe, but there was no built in support for REST based services in the IDE.

With the revealing of the new API codenamed "Indigo" the implementation of the WS* standards into an easy and coherent programming model, as well as incorporating consuming any service no matter where it was located (In process, webservice ... ) using the same programming model, WCF was born and released with .NET 3.0.

In the WCF programming model it does not matter where the service is located. The same unchanged client code can communicate with a service located in the same app domain, across app domains but in the same process, or inter process on the same machine. The same client code can also consume and use a services on different machine either on a LAN or across the inter-, intra-, extra- net.

Juval Löwy has a great note about this in his book Programming WCF Services.

Conceptually, even in C# or VB, there are endpoints: the address is the memory address of the type’s virtual table, the binding is CLR, and the contract is the interface itself. Because in classic .NET programming you never deal with addresses or bindings, you take them for granted, and you’ve probably gotten used to equating in your mind’s eye the interface (which is merely a programming construct) with all that it takes to interface with an object. The WCF endpoint is a true interface because it contains all the information required to interface with the object. In WCF, the address and the binding are not preordained and you must specify them.
Obviously due to the physical differences in how the messages are transported from client to service there are circumstances that need to taken into account. A slow WCF service located across the world, should limit the round trips as much as possible and make sure that the OperationContract is not a "chatty one". Funny enough this applies equally well for the service located in the same app domain in the same process but for a different reason. In the remote scenario it is due to performance, in the same app domain the reason is to achieve an location agnostic service, which can be moved to a remote location only by changing the endpoint  in the configuration file. This is not enforced by the framework but is good programming practices.

Now with .NET 4.0 we have built in support for calling and consuming REST based web service. However especially if you rely on http headers to transfer custom meta information to the service you will have to dive into the extensibility points of WCF. Initially this posting was supposed to be concerning that subject specifically but seeing how it ended up being more of a history lesson of WCF, I have posted that as a separate post which is located here WCF REST service with custom http header check in .NET 4
Post a Comment