"Intrapartition" refers to data on direct-access storage devices for use with one or more
programs running as separate tasks. Data directed to or from these internal
queues is referred to as intrapartition data; it must consist of variable-length
records. All intrapartition transient data destinations are held as queues
in the same VSAM data set, which is managed by CICS. An intrapartition destination
requires a resource definition containing information that locates the queue
in the intrapartition data set. Intrapartition queues can be associated with
either a terminal or an output data set. When data is written to the queue
by a user task, the queue can be used subsequently as input data by other
tasks within the CICS region. All access is sequential, governed by read and
write pointers. Once a record has been read, it cannot be read subsequently
by another task. Intrapartition data may ultimately be transmitted upon request
to the terminal or retrieved sequentially from the output data set.
Typical uses of intrapartition data include:
- Message switching
- Broadcasting
- Database access
- Routing of output to several terminals (for example, for order distribution)
- Queuing of data (for example, for assignment of order numbers or priority
by arrival)
- Data collection (for example, for batched input from 2780 Data Transmission
Terminals)
There are three types of intrapartition transient data queue:
- Non-recoverable
Non-recoverable
intrapartition transient data queues are recovered only on a warm start of CICS®. If a unit of work (UOW) updates a non-recoverable intrapartition queue
and subsequently backs out the updates, the updates made to the queue are not backed out.
- Physically recoverable
Physically
recoverable intrapartition transient data queues are recovered on warm and
emergency restarts. If a UOW updates a physically recoverable intrapartition
queue and subsequently backs out the updates, the updates made to the queue are not backed out.
- Logically recoverable
Logically
recoverable intrapartition transient data queues are recovered on warm and
emergency restarts. If a UOW updates a logically recoverable intrapartition
queue and subsequently backs out the changes it has made, the changes made
to the queue are also backed out. On a warm or an emergency restart, the committed
state of a logically recoverable intrapartition queue is recovered. In-flight
UOWs are ignored.
If an application is trying to issue a read, write, or
delete request and suffers an indoubt failure, it may receive a LOCKED response
if WAIT(YES) and WAITACTION(REJECT) are specified in the queue definition.
[[ Contents Previous Page | Next Page Index ]]