TMF – TRANSACTION MANAGEMENT FACILITY

TMF – TRANSACTION MANAGEMENT FACILITY

It is a subsystem which protects databases in online transaction processing (OLTP) environments.

The TMF subsystem supports both interactive and programmatic interfaces for commands and event messages. The TMF subsystem manages two major problems:

  • Database consistency.

  • Transaction concurreny.

TMFCOM is a command interface and provides interactive commands for all TMF operations. Also allowe you to manage, control and retrieve information about TMF objects such as dump files, and subsystem itslef.

TMF concepts:

  • Transactions – A transaction is an explicitly delimited operation or set of related operations that alters the content of a database. The range of transaction is identified within application program by a pair of statements:
    BEGIN WORK and COMMIT WORK in Non Stop SQL/MP environment.
    BEGIN-TRANSACTION and END-TRANSACTION in other programming environemnts.
    All operations within the range of a transaction are treated by the TMF subsystem as a single unit: either all of the changes performed by a transaction are ade permanent (the transaction is committed) or none of the changes are made permanent (the transaction is borted). If a failure occurs during the execution of a transaction, whatever partial changes were made to the database are backed out, leaving the database in a consistent state.

  • Audit Trail – An Audit trail is a series of files containing audit records and TMF control records. Before a transaction permanently commits it changes to the database, information about the affected database rows or records is written to the audit trail.

  • The Master Audit Trail – There is one MAT in each TMF subsystem. The MAT contains all of the TMF control records and usually also contains audit records for a set of data volumes specified by the TMF system manager.

TMF recovery processes:

  • Backout – It undoes the effects of aborted transactions. It is created when you start the TMF subsystem and automatically invoked whenever a transaction aborts.

  • Volume Recovery – This process recovers data volumes that had one or more tables or files open when a media or system failure occured.

  • File Recovery – This process uses online dumps(archived copies of database tables and files) and audit dumps (archived copies of audit trails) to restore tables and files to a consistent state after an accidental purge or a media failure.

Leave a Reply

Your email address will not be published. Required fields are marked *