There are special considerations for journal, state, and control
data sets when you work with master data sets in CXRC. In a CXRC environment,
each SDM uses its own journal, state, and control data sets. Place
these data sets so a performance bottleneck is not created due to
contention between the multiple SDMs for the same channel path, disk
subsystem, or volume resources.
Example: Suppose you have four primary disk
subsystems, generating 100 MB per second update rate and two separate
SDMs, each managing two disk subsystems. By placing the journal data
sets for both SDMs on a common disk subsystem, that subsystem will
experience a 400 MB per second update rate. The channel paths, disk
subsystem, and volumes must be able to manage this update rate.
Guidelines: Consider the following guidelines
when using journal, state, and control data sets in a coupled environment:
- Allocate the journal, control, and state data sets for the individual
sessions in the same manner as the existing support.
- Add a new member to the state data set to indicate that the session
is coupled and to allow the XSTART command to process the XCOUPLE
ADD command automatically during restart. On a new start, this member
is used to uncouple the session.
Note: The size of this
new state data set member is small and does not affect the size requirements
for the state data set.
- Define from eight to 16 journal data sets to improve the efficiency
of XRC in a large coupled environment.
- Allocate the control data set using physical sequential allocation.
Note: The state data set is updated whenever you issue XCOUPLE
ADD, XCOUPLE DELETE, and XCOUPLE PURGE commands.
If there is insufficient space in the state data set, you will
not be able to couple an XRC session to a master session. To allocate
a new, larger state data set, perform the following steps:
- Issue XSUSPEND msession_id TIMEOUT(hh.mm.ss).
- Copy all members from the existing state data set to the new,
larger data set.
- Rename the existing state data set to an unused name.
- Rename the new, larger data set to the name of the pre-existing
state data set.
- Issue the XSTART command and resume XRC operations.