Looking for advice with resources and link relations

Hey all,

I'm looking for some ideas on how to model the resources and link relations
for a Dynamic Distributed Service Locator where services register in a
directory with information like:

- Service Name
- Service Version (multiple versions of the service can exist in the
directory)
- Data Center Location
- Service Endpoint address
- Service Quality (multiple instances of the service may exist, some on
better hardware than others)

A service will register itself with the directory but also query it for
dependencies. ServiceA may require to locate ServiceB using version 2.0 but
here's where I start to get confused how I might model this, it may also
decide it prefers data center locality as a primary preference and service
quality as a secondary preference.

Administration UI's may also do broader queries to show a dashboard.

My first thoughts was something like this when ServiceA queries for
ServiceB.

/services/ServiceB/2.0?prefer=datacenter,quality

This would list all versions of ServiceB.
/services/ServiceB

Would a good idea then be to have something like this for querying what
services exist in a data center?
/DataCenterA

And maybe to get a specific version of an instance from a data center.
/DataCenterA/ServiceB/2.0

With the results returning multiple listings, I would use links so that the
client can choose the right one to use, and if its unavailable it could
fallback using another link.

Any feedback is appreciated.