Hi!
As far as I understand doing REST with XML or JSON is fine.
The URL is usually used as a resource identifier and there may be additional parameters, but they are usually used to, for instance, limit the number of results in the response (if it is a list), limit the depth of the representation etc.
Example:
http://acme.com/Customers/1123 This URL identifies the Customer with id 1123.
The representation of the resource is enclosed as a payload of the request; for instance if I am creating a new Customer, I use a POST request with the representation of the new Customer in XML or JSON. If I am retrieving a Customer, I use a request to the URL identifying the Customer and obtain a representation in XML or JSON as response.
As said before, I use Grizzly in order to make the RESTlets server cope better under load; with a high frequency of requests, there tend to be occasional losses if Grizzly is not used. Using Grizzly, these occasional losses disappear.
I'll see if I have time to show how to process payloads of request but cannot promise anything.
Best wishes!