Release 10.1A: OpenEdge Getting Started:
Database Essentials

Primary recovery (before-image) information

On systems where updates are done frequently, it is important to make the read and write access to this database area as efficient as possible. The write access is more important.

The following sections provide simple tips to create an efficient environment for the primary recovery area.

Extent size rules

These rules apply to both the primary recovery area and the data areas:

Do not make them too large.
Always make the last extent variable length.
Enable large files as a safety valve.

Enabling large files is particularly important on the primary recovery area because this is the place you are most likely to experience issues. A large update program with poor transaction scoping or a transaction held open by the application for a long period of time can cause abnormal growth of this area. If the fixed portion of an area is 2GB in size, you start extending your variable portion in that same transaction. Only then do you notice that you might need more than 2GB of recovery area to undo the transaction. If you are large-file enabled and have enough disk space, there is no problem. If you are not large-file enabled, the database might crash and not be recoverable because there is no way to extend the amount of space for the recovery area without going through a proper shutdown of the database.

Sequential access

The primary recovery area is sequentially accessed. Items are written to and read from this area in a generally linear fashion. If you are able to isolate a database’s primary recovery area from other database files and other databases, then it is a good idea to store the extents for this area on a single disk (mirror).

While striping increases the throughput potential of a file system, it is particularly effective for random I/O. If the area is not isolated from the database or you are storing the primary recovery areas of several databases on the same disk, use striping because the I/O will be fairly randomized across the databases.

BI grow option

The BI grow option to PROUTIL allows you to preformat BI clusters before the user enters the database. Preformatting allows you to write more recovery notes without needing to format new clusters while the database is online. Formatting clusters online can have a negative impact on performance. Your database must be shut down for you to grow the BI file (primary recovery area). For more information, see OpenEdge Data Management: Database Administration .