While implementing SOA and OSB, how worth is to consider change from dblink to web service?

[Deme Carva]

Where I work, there are a high number of dblink between databases. Some dblinks aim to be the way for loading immense amount of data while others only exist for simple queries.
I am wondering whether it could be better to use web services than dblink. As a principle, I am thinking that simple queries could be fit as web service to improve the reuse and governance but I find web service not apropriated for data loads.
Obviously, the decision is more complex than only thinking about tecnology. Even so, it will be helpfull to hear if there are people around the world thinking to change from dblinks to web services even heavy load as well querys with small amount of data to be trafficked and what the principles, patterns or motivations behind that decision.
Basically we are implementing SOA and OSB and I am trying the best I can to rethink how we are used to do until now. I heard from a friend of mine that we must avoid the use of dblinks and only use web services. This sounds strange to me because I am wondering how much overhead by the net we are going to face. Could someone give your personal opinion or suggest some article? Are there some advantages and disadvantages previously known about this consideration?

[Roger Sterling]

Abstracting database operations into a Web Service is an ideal solution as long as the Web Service can support the throughput and security necessary. Some tables have very specific restrictions on who can access what data. Your Web Service abstraction must pass these operating parameters along to the caller so the caller does not gain access to resources the caller should not have access to. You could query the people over at Enterprise Service Bus Q&A for more ideas if you still need help with this. There is a dedicated forum for OSB.