In these systems, the .idx file contains a sorted list of keys and pointers (often 64-bit or 128-bit addresses). When a user queries a record, the software queries the .idx file, traverses the tree structure, retrieves the memory offset, and performs a direct seek operation on the primary data file (often denoted as .dat or .db ). This architecture allows for rapid rebuilding of database integrity; if the data file remains intact, a corrupt .idx file can often be regenerated without data loss.