When a transaction or command (including utilities) is executed, one or more of the following phenomena might occur, causing communication errors. This might result in a transaction or command error, or a unit abnormality might be detected by host-to-host monitoring (the KFPS05289-E message is output).
Should such phenomena occur, reduce the communications load using the following procedures as applicable.
When you execute a pdload command, create input files for each partition storage condition to reduce the communications load when the pdload command is executed in RDAREA units.
You can also reduce the communications load by placing the multiple input files that have been created not in one place (on the same machine), but on a server machine that has a table storage RDAREA.
When you reorganize a table, unload a table or reload a table using the pdrorg command, execute the command in RDAREA or server units to reduce the communications load.
When you re-create indexes, reorganize indexes, or batch create indexes using the pdrorg command, execute the command in index or server units to reduce the communications load.
When you execute the pdcopy or pdrstr command, reduce the communications load as follows.
You can also reduce the communications load by placing backup files on the server machine that processes commands.
Since accessing data on multiple back-end servers through a single transaction generates data communication between servers, you need to reduce data communication routes between servers as much as possible. Data on multiple back-end servers is accessed when any one of the following conditions is met.
When any one of the above conditions is met, you can reduce the data communication routes between servers by taking the following actions.
The communications load increases when floating servers are used, so avoid floating servers whenever possible. Note that floating servers are used when any one of the following conditions is met:
When any one of the above conditions is met, you can reduce the number of floating servers used by taking the following actions.
For details about SQL optimization options, see the HiRDB Version 9 UAP Development Guide.
When shared tables are updated through multiple front-end servers, the communications load increases. For this reason, when you update multiple shared tables, connect the HiRDB client to the same front-end server whenever possible.
When a system has many back-end servers, utilities might terminate abnormally when many of them are executed simultaneously. Should this occur, take a corrective action such as reducing the number of utilities that execute simultaneously.