Working around with Alfresco document management systems, it comes with an architecture to handle and dealing with files likely, in terms on building your interfaces that request/response the petition of document to be indexing, profiling, storing and managing following CMIS specification, the best way of doing an implementation of an interface if your using Web Service is handle the archive as attachments or MTOM in the best way previously passing the file thought discovering, encoding, binarization and packing for sending a file to well decoupled layer of services that handle asynchronously the reception of document and retrieving a correct response to a client who is waiting with a timeout in queue/streaming manner the processing of his document.
Alfresco uses open source in backstage (Apache Chemistry) which intermediately use a language to communicate repository with internal database is like JPQL or LINQ, is just simple like using hibernate HQL and you must develop an integration to dealing the document structure and normalization of their characterization with your systems, so to dealing with connection you could query Alfresco with: Chemestry OpenCMIS Workbench to start the tool you must execute:
$ workbench.sh
and you will get like this in Java Application:
Commons problems in development are that tables with column ObjectId respond lower in amount of time of 20 seconds after 2 million of records saved in the documents, this sound like a tunning database problem, solving will require some investing hours, after taking care of network problems handling with High Availability in your infrastructure and asynchronously retrieve a response of events and states with steps of document process.
Best Regards,
