Since information is often a primary target in a computer crime, organizations that store their information in database management systems (DBMSs) must develop a capability to perform database forensics. This paper describes a database forensic method that transforms a DBMS into the required state for a database forensic investigation. The method segments a DBMS into four abstract layers that separate the various levels of DBMS metadata and data. A forensic investigator can then analyze each layer for evidence of malicious activity. Tests performed on a compromised PostgreSQL DBMS demonstrate that the segmentation method provides a means for extracting the compromised DBMS components.