URI best practices

Is there any recommendation for URI schema for the following use case: We have a few separate web services providing RDF, however it doesn't look as a good idea to use the web service URL as the data URI. Meanwhile we would like to maintain the web experience, so you can download usefully information from the URI itself, which is not the service. Our assumption is to enable redirects on the URIs pointing to the services to provide useful information.

  1. Is this a recommended architecture?
  2. Is there any documentation on a similar implementation we can start with?
  3. Is there any best practice for the URL structure? Like http://data.example.com/... or http://example.com/data/... or... ? Think about a busy existing website on example.com

I believe the following document provides recommendations on how to address use cases analogous to yours.

http://www.w3.org/TR/cooluris/

Tom Heath and Christian Bizer's book "Linked Data: Evolving the Web into a Global Data Space" is good on issues like this. It's available online at http://linkeddatabook.com/editions/1.0/ . One part that discusses the issue you raise is titled "Abstract away from implementation details".

My first recommendation would be to expose a SPARQL endpoint.

Secondly, you might want to consider proxying your other services through this endpoint variably (try not to do it in too crufty a way).

Thirdly, read the document mentioned by @beroca since it is the answer to your second question and will heavily influence the third.

You should then be in a position to come back with more specific questions.