Clustered file organization

A clustered file is also called a B-tree-clustered file is a tree-structured file in which records with adjacent index values are clustered together to reduce the cost of searching for ranges of records. The SFS server voluntarily maintains the clustered file organization used in clustered SFS files.

Records in a clustered file don't have a numeric record index such as an ESN or RSN. The primary index is able to be based on any field or combination of fields and the records in a clustered file are ordered based on the contents of the primary index. Ever since the SFS server is able to move records to maintain clustering when new records are inserted or deleted there is no practical way to maintain direct references to individual records. The records in a clustered file are able to be fixed or variable in length. Records in clustered files are able to be updated or deleted. Disk space freed by means of record deletions is automatically reused. It works with blocks of data that are exchanged with disk or tape systems.