Release 10.1A: OpenEdge Data Management:
DataServer for Microsoft SQL Server

Monitoring cursor and connection use

Monitor the use of your managed connections to tune the number of connections you allocate. If you regularly exceed your allocation, consider increasing the number of managed connections. If you never use your total allocation, consider decreasing the number of managed connections.

Connections

The DataServer log contains messages indicating the status of connections. At startup, the number of managed connections initialized is written to the log file. If connections are rejected, this is also logged. If a connection from the ODBC connection pool cannot be reused, the ODBC driver issues the message DEAD Connection which is written to the log file.

At the end of a session, the DataServer writes summary information about cursor use. The summary contains the following information:

Needed connections — The number of connections actually used during a session.
Peak connections — The maximum number of connections used simultaneously during a session.

When debug logging is enabled with the “-Dsrv qt_debug,cursor” switch, a summary of connection activity is written to the log file. This summary contains:

Number of connections (defaults to 5).
Number of peak connections (defaults to 5).
Highest connection value.
Lowest connection value.
Number of connections pulled out of pool.
Number of connections placed in the pool.
Total sum of active connections.

The following excerpt from a dataserv.lg file shows the summary of the managed connection activity written when the “-Dsrv qt_debug,cursor” switch is used, plus the standard messages:

16:32:09 : Standard secondary connection: 16:32:09 : Connection chain 0xbc6e64: num_conns 1 16:32:09 : Connection chain 0xbc6e64: peak_conns 1 16:32:09 : Connection chain 0xbc6e64: high_conns 1 16:32:09 : Connection chain 0xbc6e64: low_conns 0 16:32:09 : Connection chain 0xbc6e64: num_off_conns 1 16:32:09 : Connection chain 0xbc6e64: num_on_conns 1 16:32:09 : Connection chain 0xbc6e64: sum_active 1 16:32:09 : Connection chain 0xbc6e64: Needed Connections: 1, (Needed 4 less than requested) 16:32:09 : Connection chain 0xbc6e64: Peak Connections: 1, (Peaked at 4 less than requested) 16:32:09 : Connect: 0x1012398 Stmt: 0x1b95f80 Handle: 0x1011d98 Crc: 1613 Proprty: Statement dropped 16:32:09 : Statement cache (-Dsrv PRGRS_PREPCACHE setting) reuse ratio is 0.

For more information on using the qt_debug option to the -Dsrv switch, see Appendix B "Server Related Command Line Utilities and Startup Parameters."

Cursor downgrades

Cursor downgrades are monitored by the DataServer, according to the type of cursors in use:

For firehose cursors, the DataServer counts the number of times a cursor is downgraded, and writes the count to the log file at the end of the session.

Consider increasing the number of DataServer managed connections if firehose cursors are frequently downgraded.

For downgrades of Fast Forward-Only cursors, a message is written to the log file when the downgrade occurs.

Statement Cache

The reuse of the statement cache is calculated as a ratio. Firehose cursors decrease the probability of a statement being reused. A reuse ratio of 3 indicates that statements are reused an average of 3 times. Reuse ratios of 3 or higher indicate good utilization of the statement cache. Reuse ratios less than 1 indicate poor utilization of the statement cache. If your ratio is less than 1, consider disabling the statement cache with “-Dsrv PRGRS_PREPCACHE,0”. For more information on the PRGRS_PREPCACHE option to the -Dsrv switch, see Appendix B "Server Related Command Line Utilities and Startup Parameters.".