There's lots of good information and guidelines in the NISO report entitled Best Practices for Designing Web Services (PDF) in the Library Context. Much of the document is general, and applicable to most disciplines. Richard mentioned it in his recent post. What I don't see in the document is much mention of SOA, except the glossary entry, and the IBM definition in the forward. You don't need SOA to do web services... but web services without SOA is done at your peril.
Appendix A provides some guiding examples of categories for library web services. We've been on the SOA and web services path at CISTI for awhile now, and indeed most of our services fit into this general framework, but it's nice to confirm that fact, and to consider the problems in this new context.
The document outlines several types of services for consideration for the library:
- Discovery services - Discover metadata, full text or other services.
- Locate services - Resolve an object to its location
- Requesting services - I'd include book loan, or document order requests here perhaps.
- Delivery services - conversion services, information/electronic document delivery services perhaps
- Common services - backoffice, financial and all the usual suspects.
Too often in the library context we've mixed functions together, much to our peril. Discover and locate are entirely different functions. In moving forward, we can't assume that because we have metadata concerning an item, that we actually hold a particular item... nor can we assume that because we don't hold and item that it's not retrievable for the client.
It will be interesting to map our existing and planned services in this context.