RDF System Management Manual for H-Series RVUs (RDF 1.8)

RDF Subsystem #2
\A ---------> \C
Because the two subsystems run independently of one another, if system \A fails and you execute
TAKEOVER commands on systems \B and \C, the two backup databases might not be
synchronized with one another. The extractor for the \A-to-\B subsystem, for example, might
have replicated audit data to system \B, but, before the extractor for the \A-to-\C subsystem
could replicate the same data to system \C, system \A failed. To correct this situation, you issue
a COPYAUDIT command to transfer the extra audit data from system \B to system \C. You
then reissue the TAKEOVER command on system \C, and the two backup databases are logically
identical. At this point you can then continue application processing from system \B to system
\C or from system \C to system \B within minutes of losing system \A.
Using ZLT to Achieve the Same Protection
To achieve the same result using the RDF/ZLT product you configure system \B as the ZLT
standby node for both RDF subsystems #1 and #2. Upon losing system \A, you connect the
remote mirrors to system \B and issue TAKEOVER commands on both systems \B and \C.
Since, as part of the ZLT takeover, each subsystem fetches the final audit data from the remote
mirrors connected to system \B, both backup databases receive the same data. When the takeover
operations are complete, the databases on systems \B and \C are logically identical to one
another, and you have achieved that without executing a COPYAUDIT command (and you also
have not lost any committed data).
Summary
To be able to use the triple contingency feature, you must:
1. Establish two RDF configurations with the same primary system and separate backup
systems.
2. Ensure that the hardware configurations of the two backup systems are identical with regard
to data volumes and image trail volumes.
3. Ensure that the data volumes and image trails of the two RDF configurations are configured
identically with respect to the two backup systems (with the few minor exceptions noted
earlier in this chapter).
4. Set an adequate purger RETAINCOUNT parameter on the backup systems (it must be the
same on both).
Upon loss of the primary system, you must:
1. Issue a TAKEOVER command on both backup systems.
2. When the takeovers have completed successfully, examine the EMS event log on both backup
systems for a 735 message to determine which system is missing audit information.
3. On the system with the least amount of audit information, issue a COPYAUDIT command
specifying the name of the other backup system and its RDF control subvolume.
4. When the COPYAUDIT command has completed successfully, issue a second TAKEOVER
command on that same system.
5. Initialize, configure, and start RDF on whichever system you want to be the primary in the
new configuration.
6. Start application processing on the new primary system.
260 Triple Contingency