When Distributed Data Processing is NOT Overridden
The following considerations apply for opening distributed files when distributed
data processing is not overridden:
The system will process a distributed file that is open for input-only
using buffered retrieval (*BUFFERED). Buffered retrieval will achieve the
best performance along with maximum record concurrency, however, you might
not see all of the changes made to the file as they occur. Refer to Input/Output Considerations for Distributed Files for more information.
The system will process a distributed file that is opened for output-only
one record at a time. If your distributed file is opened for output-only,
the DSTDTA parameter will have no effect. Also, if SEQONLY(*YES) processing
has been requested, it will be changed to SEQONLY(*NO). The SEQONLY(*NO) processing
will provide feedback on a record-by-record basis when the records are inserted
into the file.
The system will process a distributed file that has been opened with an
option that includes update or delete using non-buffered retrieval (*CURRENT).
Non-buffered retrieval ensures that you are updating or deleting the same
record that would have been updated or deleted if all of the distributed file
data had been contained in a non-distributed database file. Since non-buffered
retrieval will be used, the best performance for the distributed file will
not be achieved, but the best data integrity and the maximum record concurrency
will be guaranteed.
Note:
For arrival sequence distributed files,
records will be retrieved in arrival sequence starting with the first node,
then the second node, and so on. For duplicate key considerations, refer to Input/Output Considerations for Distributed Files.
The system will process a distributed file that is opened with all operations
(*INP, *OUT, *UPD, *DLT) using non-buffered retrieval (*CURRENT), since it
includes both update and delete options.
(C) Copyright IBM Corporation 1992, 2006. All Rights Reserved.