You are correct, it sounds like an ESP is what you want. In general, an ESP works seamlessly behind the scenes. The UI for interacting with Jive is the same; however, on the backend ... binary is transmitted via middleware to the storage destination, and made available on-demand in similar fashion. To be 100% transparent, ESF is a very involved integration ... and you'd definitely be better getting an integration off-the-shelf or from an experienced partner.
2.) Yes, Search can work on external documents. In this scenario, maintains a local index of the searchable content/metadata from every document that Jive incorporates into it's search function.
There are solutions available from Egnyte, Box, Google Drive, Office 365, etc... check out this page for more details:
Hope that helps. =)