Introduction to Data Management

Protecting Databases With RDF
8-6 15873 Tandem Computers Incorporated
Primary System Processes
At the primary site, RDF processes monitor the TMF audit trail files associated with
the primary database and extract information from them, filtering out before-images
and other irrelevant records. TMF after-images showing the state of the database
records after any update are immediately extracted from the audit trail files and sent
over a communications line to an RDF process at the backup site.
There are two RDF processes on the primary node: one monitors the audit trails and
controls all RDF processes, and the other extracts information from the audit trails and
transmits that information.
Backup System Processes
When an RDF process at the backup site receives the audit records, it stores them in
temporary disk buffer areas. Then other RDF processes on this node can use a
continuous rollforward mechanism to apply the after-image values to the
corresponding records in the secondary database. In this way, the secondary copy of
the database is usually current within seconds to minutes of the primary database.
There are two RDF processes on the backup node; one receives the transmitted data
from the primary node, and the other applies the data to the backup database.
Using RDF RDF runs on NonStop systems with the NonStop II, TXP, VLX, EXT, and CLX
processors, using the DP2 disk process. RDF transmits data on EXPAND
communication lines only. RDF requires you to make sure that your system has this
general hardware and software environment in which to operate, and that it meets
other specific requirements involving such factors as audited files, maximum number
of volumes per system and files per volume. Then you can perform the following
general tasks:
1. Install RDF on both your primary and backup systems.
2. Create the RDF configuration file, using RDFCOM from your primary system.
Here, you specify global parameters that will apply to all of RDF plus specific
parameters that will apply to the specialized processes that execute within RDF.
3. Back up the primary database from the primary system, and restore it to the
secondary system.
4. Initialize and start TMF on the primary system.
5. Initialize and start RDF on the primary and backup systems, using RDFCOM
commands. This step starts the execution of the RDF processes.
Now, the data at the primary site is covered by full RDF protection. If the primary
system fails, you switch application processing to the backup system and its database
by issuing a single RDFCOM command from your backup system.