Organizations have numerous operational systems, each with its own dedicated business function, tailored information structure, analytics, and reports. Secondary aggregations of data are common, and include such things as data warehouses, operational data stores, and data marts. There are countless information silos, each particular to its mission or function.
Traditional federated search systems involve the user querying the database of each silo for relevant content. More sophisticated federated search systems use intelligent middleware to broker the individual queries to each database, a model where the middleware processes the query by engaging the myriad of information silos automatically and compiling the findings, returning the collective results to the inquirer.
Federated search assembles cross-silo data "just-in-time," at the point information is needed. Although this type of federated search is applicable in some settings, it is not well suited to high-performance enterprise discoverability, which is required to deliver on data finds data.
There are two primary reasons federated search does not scale:
Existing systems generally do not have the indexes necessary to enable the efficient location of a record. Payroll systems, for example, will often have prebuilt indexes (defined pointers into the data) to facilitate searches on employee number, tax ID number, and name. Rarely would a payroll system have an efficient way to locate records on address ...