RDF/IMP and IMPX System Management Manual (RDF 1.3+)

Managing RDF

Compaq NonStop™ RDF/IMP and IMPX System Management Manual—522204-001

5-15

Standard Configurations

In a standard RDF configuration (system \A the primary, system \B the backup), the

steps for performing a planned switchover from \A to \B are:

1. On system \A, stop the business applications that access the primary database.

2. On system \A, issue a STOP TMF command.

TMF stops as soon as all outstanding database transactions are either committed or

aborted. It then writes a shutdown record to the MAT. Subsequently, the RDF

subsystem shuts down when all updater processes on the backup system (\B) have

reached the shutdown record in the image trail file. At this point, the primary and

backup databases are identical.

3. On system \B, note the local system time; you’ll need it later.

4. On system \B, restart the business applications.

At this point, the RDF subsystem is stopped, the business applications from system

\A are running on system \B, and all audit information is being queued in TMF audit

trails on system \B.

5. When system \A is ready to resume its normal operations, restart TMF on \A.

6. On system \B, issue an INITIALIZE RDF command using the INITTIME option

and specifying the local system time you noted in step 3.

This action initializes the RDF extractor on \B to begin reading the MAT at the first

record after the TMF shutdown record that was generated in step 3. Because the

shutdown record was created before the business applications generated any updates

on \B, the extractor cannot miss any relevant audit information.

7. On system \B, configure the RDF subsystem to run from \B to \A.

8. On system \B, start the RDF subsystem. The RDF subsystem begins replicating

database changes from \B to \A.

When the extractor on \B starts sending current audit information to \A (and at a

convenient time with respect to your processing environment), stop the applications on

\B and carry out a planned switchover from \B to \A.

The planned switchover repeats the procedure described above, except that you reverse

the roles of systems \A and \B. After doing so, RDF replication once again occurs from

\A to \B.