Built on top of COM are core facilities for storing and naming objects and for transferring data between them. Together, these facilities form the core of system-wide information management. You can take advantage of these facilities by using OLE's supplied implementations, or you can extend or replace them with your own. These core facilities include structured storage, monikers, and uniform data transfer. Together they provide the infrastructure required for objects to interrelate across process, machine, and network boundaries.
OLE's structured storage model defines an architecture for storing and retrieving objects that reside inside files or other containers. The basic problem addressed by the structured storage model is how to write an object to and retrieve an object from any persistent storage when that object resides inside the flat file space of some container. OLE solves the problem by providing interfaces that support storage and stream objects. A storage object provides a hierarchical structure for mapping the location of storage and stream objects just as directories provide a way of locating files on disk. A stream object contains the object's data, or contents. Using structured storage is mandatory if you want to support compound documents or other forms of containment. OLE also provides a default implementation of structured storage called compound files.
OLE supports the persistent naming of objects through a mechanism called a moniker. A moniker provides a way for clients of an object to locate that object even when it resides in a different process or on a different computer. Objects have a file moniker, which is similar to their absolute pathnames, but their clients normally bind to them using a relative moniker, which indicates the object's location relative to the client. In this way, if client and object are both relocated, they can maintain their connection as long as their locations relative to each other remain the same. In general, a moniker also enables clients to locate and connect to some portion of an object — a certain range of cells in a spreadsheet, for example — by providing an item moniker, by which that pseudo-object is always known.
OLE's uniform data transfer model provides a single, standard mechanism for transferring objects and data between applications. Through a single interface, OLE supports standard Microsoft® Windows® clipboard transfers, as well as drag and drop, and embedded objects. Uniform data transfer greatly enhances the clipboard model by providing rich structures for describing the format of the data to be transferred and the storage medium in which a transfer is to occur. For example, instead of always having to pass data in a global memory handle, you can choose from among several storage media, including a disk-based file and pointers to OLE storage and stream interfaces.
Built on top of COM, structured storage, monikers, and uniform data transfer are the three technologies for which OLE is best known: compound documents, Automation, and OLE controls. For many software vendors, the desire to incorporate one or both of these technologies in their applications, or to make sure their applications work easily with operating systems and other applications that do, has been the prime motivation for turning to OLE.