From PIT (Practical IT) Wiki
Jump to navigation Jump to search

HADR is the acronym for High-Availability and Desaster Recovery. Several DB2 machines are clustered in order to achieve these goals.

At least two machines will be needed to create an HADR cluster.


The primary Server is contacted by applications, a direct connect to a StandBy Server is not recommended in case his role as STANDBY should be feasible. Only the PRIMARY connects to that machine and provides changes in the database via archive logs.

In case of a failover on the PRIMARY the roles can be switches automatically and manually so even the a machine burns down the other machine will persist and secure the current data stored on it.

-> It makes sense to put the DB Servers in different locations, doesn't it?

If more than one STANDB server is needed (e.g. as read-only server for statistical applications like DataWarehouse, or SAS the role of those Servers will need amendments:

  • The REAL STANDBY will get the PRINCIPAL STANDBY role.
  • All additional STANDBY machines take the role as AUXILIARY STANDBY.

The SYNCMODE of the database should be reviewed to meet the needs in the HADR situation.

For more Information please consult the HADR wiki linked above.



YouTube: Ember Crooks on HADR

PDF: Ember Crooks on HADR