The solidDB AuditInfo feature generates a set of human readable files in the file system. The files contain a full set of records on all SQL executions, data model operations, user logins, HotStandby state changes, and other operations that can be monitored.
To analyze database operations by using the contents of the audit log files requires work (or a separate analysis program), and an understanding of the structure of the files. However, solidDB does not include any analysis tools.
As the files are in human readable format, it is strongly recommended that access to the directory that contains the files should be strictly controlled.
Audited operations
To set up audit policies, you must understand the type of information that the AuditInfo feature can and cannot collect.
The solidDB AuditInfo feature can collect the following information:
▪ All SQL operations (DML, DDL and ADMIN COMMANDS) that are executed by using ODBC and JDBC interfaces
▪ All SQL operations that are executed inside solidDB stored procedures and triggers
▪ Operational events that are necessary to ensure the completeness of audit information such as startups, shutdowns, connects, disconnects, transaction commits and rollbacks, and HotStandby role changes
▪ SQL operations executed by solidDB tools that use ODBC to execute SQL (solloado)
The solidDB AuditInfo feature cannot collect the following information:
▪ Data operations executed by using the SA interface, including SA-based solidDB tools (solload).
▪ Operations where data is replicated to the server from another server. For more information about these technologies, see Replication.
▪ Triggers that modify column values inside trigger code. However, statements that are executed inside the trigger code are treated as regular statements and can be audited.
▪ Information about user passwords. The statements that impact passwords (CREATE USER and ALTER USER) have the password part scrambled.
Note The solidDB AuditInfo feature might not display long BLOB data completely. There is a limit after which the BLOB information is truncated. Also, if BLOB data is written as an SQL literal, a long SQL statement might be truncated.