Normally, the ARCRET utility retrieves transactions interactively while sequentially accessing your archive index systems. In some instances, however, the underlying archive system does not support retrieving transactions while sequentially processing the records.
This is the case with Documanage. Since Documanage serves as both the archive index and storage system, it is cannot currently be accessed in the same manner as a database. Therefore, use the /BQ option to queue sequential transactions in memory before attempting to retrieve the associated files.
Queuing the transactions this way consumes more memory than would otherwise be required. In general, you can determine the index record size by examining the APPIDX.DFD file you use. Add some extra for overhead and then divide that into the maximum amount of memory that you wish to consume—over and above that used by the ARCRET utility itself. To determine this, use the Windows NT Task Manager to get an idea of how much memory is in use while you run the utility on a single transaction.
Batch queuing also affects performance because the utility has to reset the sequential index search between each set of records. For instance, suppose it reads and queues 100 transactions in a set. After the set is processed, it has to re-read sequentially from the beginning of the index back down to the 100th record to prepare for the next set. For each set of transactions, this resetting takes longer and longer, because it has to start at the beginning of the index each time.
Note | Only use the /BQ parameter when the source archive system that ARCRET utility is reading does not support the interactive retrieval of documents while sequential reading the index records. |
© Copyright 2012, Oracle and/or its affiliates. All rights reserved. Legal notices.