Routers

Marshal implements a general harvester, the TCP JSON Harvester, that allows external applications to implement the harvesting process and act as if they where an integral part of Marshal. These external harvesters, referred to as routers, allows browsing the data source and continuous, asynchronous, data retrieval, i.e. they provide functionality equal to that of the ODBC harvester. This means that Routers mimic relational databases in the sense that the returned data is represented as table-rows with columns.

The image above shows how a router works at a conceptual level. It simply routes the Marshal harvester to the data source, hiding any and all data source specific details. 

The purpose of the router mechanism is to accomplish code separation in a platform independent way. However, it also opens for distribution of the harvesting software across multiple computers. When deploying the router on a computer other than the one running Marshal (Archiver or Editor) make sure you've properly secured access to it.

All routers share the same set of configuration properties, see below. The interpretation of most of these settings is left to the specific router. The Host and Port settings, however, have a common meaning since they are used by Marshal to connect to the router.