Journal caching can be used in replication environments as well as by journals used internally by MIMIX. Journal caching is an attribute of the journal that is defined in the journal definition. When journal caching is enabled, the system caches journal entries and their corresponding database records into main storage and writes to disk only as necessary. This means that neither the journal entries nor their corresponding database records are written to disk until an efficient disk write can be scheduled. This usually occurs when the buffer is full or at the first commit, close, or force end of data. Because most database transactions must no longer wait for a synchronous write of the journal entries to disk, the performance gain can be significant.
For example, batch operations must usually wait for each new journal entry to be written to disk. Journal caching can be helpful during batch operations when large numbers of add, update, and delete operations against journaled objects are performed.
For more information about journal caching, see IBM’s Redbooks Technote””Journal Caching: Understanding the Risk of Data Loss”.