At the heart of the Generic Search Server is the concept of searching an Index, which is a performant, non-database representation of a set of related searchable data. A Generic Search Server Index is an "inverted index" that maps words to database records that they appear in.
When searching an Index for a word, all matching records are retrieved without having to search large datasets. As a result, such Indices scale well, and for large systems it will be possible to run multiple Indices in parallel, allowing for excellent search performance if the right deployment configuration and Index tuning parameters are chosen.
Developers creating application searches do not manipulate or maintain Indices directly - all of this is handled for them behind the scenes by the Generic Search Server.