Auditing
This article explains how to use the auditing feature in Exasol.
Exasol allows you to have auditing for sessions and SQL command execution. Auditing does not require any transaction log or any database schema modification, and therefore has no performance impact on the system.
Auditing captures the following:
- All sessions with information such as driver, client, host, and operating system user
- All executed SQL statements with information such as CPU, network, and hard disk read/write
Capturing the above information facilitates debugging and data security. For example, by looking at the audit logs you can check which query used a high amount of temp DB RAM at any time. You can also see which user dropped a table at what time.
Implementation
- SaaS databases are created with auditing enabled by default.
- The audit information is logged into the following system tables:
- EXA_DBA_AUDIT_SESSIONS: The system table stores all the sessions from the moment you enable it and start the database.
- EXA_DBA_AUDIT_SQL: The system table stores all executed SQL statements from the moment you enable it and start the database.
- EXA_DBA_AUDIT_IMPERSONATION: The system table stores all of the impersonation commands that each session executes from the moment you enable auditing and start the database.
- The system tables are accessible to a user with SELECT ANY DICTIONARY system privilege.
- You can view the size of the auditing data in EXA_STATISTICS_OBJECT_SIZES.
- You can delete the audit logs by using TRUNCATE AUDIT LOGS command.
Examples
Auditing for a user
Here is an example of auditing all sessions of user JOHN
in the last 24 hours.
SQL Statement
SELECT SESSION_ID, USER_NAME, LOGIN_TIME, LOGOUT_TIME
FROM EXA_DBA_AUDIT_SESSIONS
WHERE USER_NAME = 'JOHN'
AND CLIENT NOT LIKE '%[Meta]%'
AND add_days(LOGIN_TIME, 1) > SYSDATE;
Output
SESSION_ID | USER_NAME | LOGIN_TIME | LOGOUT_TIME |
---|---|---|---|
1460639103178893350 | JOHN | 2019-11-20 11:11:18.221 | 2019-11-20 12:00:10.22 |
1460641579837871138 | JOHN | 2019-11-21 09:23:18.141 | 2019-11-21 10:20:28.240 |
1460639332787506215 | JOHN | 2019-11-22 05:33:18.229 | 2019-11-22 07:05:10.210 |
Auditing for a session
Here is an example of auditing a session 1460639103178893350
.
SQL Statement
SELECT STMT_ID, START_TIME, STOP_TIME, SUCCESS, SQL_TEXT
FROM EXA_DBA_AUDIT_SQL
WHERE SESSION_ID = 1460639103178893350
AND STMT_ID > 6
ORDER BY 1;
Output
STMT_ID | START_TIME | STOP_TIME | SUCCESS | SQL_TEXT |
---|---|---|---|---|
7 | 52:54.5 | 52:54.6 | TRUE | create schema JOHN; |
8 | 58:47.8 | 58:47.8 | TRUE | create table tmp_customers ... |
9 | 58:55.5 | 58:55.6 | TRUE | create view ... |
10 | 59:42.5 | 59:42.5 | TRUE | commit; |
11 | 59:50.7 | 59:50.8 | FALSE | IMPORT INTO tmp_customers ... |
12 | 02:51. | 02:51.5 | TRUE | rollback; |
Delete auditing log
Here is an example of deleting the audit logs before a specified date.
SQL Statement
-- deletes all auditing information before the specified date
TRUNCATE AUDIT LOGS KEEP FROM '2019-01-01';
Best Practices
Here are some best practices to ensure the data security and better auditing:
- Enable auditing for your database.
- Use FLUSH STATISTICS statement to update the statistical tables on demand.
- Always implement a user concept that distinguishes between regular users and administrators.
- Design and develop a comprehensive rights management system.
- Grant permission to a user only if it is required.
- Limit and secure any third-party access to the system.
- Separate data and views so that users may only have access to the views.