Lets assume that I'm dealing with a service that involves sending large amounts of data.
If I implement this with WCF, will WCF throttle the service based on how much memory each request takes to serve? Or will I be getting continuous out of memory exceptions each time I receive a large number of hits to my service?
I'm quite curious as to dealing with this problem outside of WCF, I'm still a bit new to service development...
While using the binding attributes and readerQuotas like Andrew Hare suggests will allow for essentially an unlimited size for most practical uses, keep in mind that the you will run into other issues such as timeouts if you accept a long running command, no matter how that service is constructed (using WCF or not).
No matter what the size of your message is, the WCF service will need to be throttled for performance so that it is not flooded. If you are hosting it in IIS or WAS, you will have additional built-in features to those hosting environments that will make your service much more "highly available". However, you still need to pay attention to concurrency issues. The following WCF config provides an example of setting some throttling values.
<system.serviceModel>
...
<behaviors>
<serviceBehaviors>
<behavior name="GenericServiceBehavior">
<serviceTimeouts transactionTimeout="00:09:10"/>
<serviceThrottling
maxConcurrentCalls="20"
maxConcurrentSessions="20"
maxConcurrentInstances="20"
/>
</behavior>
</serviceBehaviors>
</behaviors>
</system.serviceModel>