The Adaptive Server transaction log is a special heap table that stores information about data modifications in the database. The transaction log is always a heap table; each new transaction record is appended to the end of the log. The transaction log has no indexes.
Place logs on separate physical devices from the data and index pages. Since the log is sequential, the disk head on the log device rarely needs to perform seeks, and you can maintain a high I/O rate to the log.
Transaction log writes occur frequently. Do not let them contend with other I/O in the database, which usually happens at scattered locations on the data pages.
Besides recovery, these operations read the transaction log:
Any data modification performed in deferred mode.
Triggers that contain references to the inserted and deleted tables. These tables are built from transaction log records when the tables are queried.
Transaction rollbacks.
In most cases, the transaction log pages for these queries are still available in the data cache when Adaptive Server needs to read them, and disk I/O is not required.
See “Keeping transaction logs on a separate disk” for information about how to improve the performance of the transaction log.
