Performance Considerations

The following connection options can enhance driver performance. You can also enhance performance through efficient application design. Refer to "Designing ODBC Applications for Performance Optimization" in the DataDirect Connect for ODBC and Connect XE for ODBC Reference for details.

The option names found on the tabs of the driver Setup dialog box are the same as the connection string attribute names unless otherwise noted in parentheses. The connection string attribute name does not have spaces between the words. For example, the option name Application Using Threads is equivalent to the connection string attribute name ApplicationUsingThreads.

Application Using Threads (ApplicationUsingThreads): The driver coordinates concurrent database operations (operations from different threads) by acquiring locks. Although locking prevents errors in the driver, it also decreases performance. If your application does not make ODBC calls from different threads, the driver has no reason to coordinate operations. Disabling the Application Using Threads option prevents the driver from coordinating operations and improves performance for single-threaded applications.

NOTE: If you are using a multi-threaded application, you must enable the Application Using Threads option.

Array Size (ArraySize): Values for this option can be an integer from 1 to 4,294,967,296 (4 GB); the default is 60000. The value 1 is a special value that does not define the number of bytes but, instead, causes the driver to allocate space for exactly one row of data. Larger values increase throughput by reducing the number of times the driver fetches data across the network when retrieving multiple rows. Smaller values increase response time, as there is less of a delay waiting for the server to transmit data.

For example, if your application normally fetches 200 rows, it is more efficient for the driver to fetch 200 rows at one time over the network than to fetch 50 rows at a time during four round trips over the network.

Cached Cursor Limit (CachedCursorLimit): To improve performance when your application executes concurrent Select statements, Cursor Identifiers can be cached. In this case, the Cursor Identifier is retrieved from a cache rather than being created for each connection. When an Identifier is needed, the driver takes one from its cache, if one is available, rather than creating a new one. Cached Cursor Identifiers are closed when the connection is closed. To cache Cursor Identifiers, the CachedCursorLimit option must be set to the appropriate number of concurrent open Select statements.

Cached Description Limit (CachedDescLimit): The driver can cache descriptions of Select statements and improve the performance of your ODBC application. Therefore, if your application issues a fixed set of SQL queries throughout the life of the application, the description of the query should be cached. If a description is not cached, the description must be retrieved from the server, which reduces performance. The descriptions include the number of columns and the data type, length, and scale for each column. The matching is done by an exact-text match through the From clause. If the statement contains a Union or a subquery, the driver cannot cache the description.

Catalog Functions Include Synonyms (CatalogIncludesSynonyms): Standard ODBC behavior is to include synonyms in the result set of calls to the following catalog functions: SQLProcedures, SQLStatistics, and SQLProcedureColumns. Retrieving this synonym information degrades performance. If your ODBC application does not need to return synonyms when using these catalog functions, the driver can improve performance if this option is disabled.

Catalog Options (CatalogOptions): If your application does not need to access the comments/remarks for database tables, performance of your application can be improved. In this case, the CatalogOptions option should be disabled because retrieving comments/remarks degrades performance. If this option is enabled, result column REMARKS (for the catalog functions SQLTables and SQLColumns) and the result column COLUMN_DEF (for the catalog function SQLColumns) return actual values.

Default Buffer Size for Long/LOB Columns (DefaultLongDataBuffLen): To improve performance when your application fetches images, pictures, or long text or binary data, a buffer size can be set to accommodate the maximum size of the data. The buffer size should only be large enough to accommodate the maximum amount of data retrieved. Otherwise, performance is reduced by transferring large amounts of data into an oversized buffer. If your application retrieves more than 1 MB of data, the buffer size should be increased accordingly.

Describe At Prepare (DescribeAtPrepare): When enabled, this option requires extra network traffic. If your application does not require result set information at prepare time (for instance, you request information about the result set using SQLColAttribute(s), SQLDescribeCol, SQLNumResultCols, and so forth, before calling SQLExecute on a prepared statement), you can increase performance by disabling this option.

Enable Scrollable Cursors (EnableScrollableCursors) and Enable Static Cursors for Long Data (EnableStaticCursorsForLongData): When your application uses Static or Keyset (Scrollable) cursors, Enable Scrollable Cursors must be enabled. Also, if your application retrieves images, pictures, long text or binary data while using Static cursors, Enable Static Cursors For Long Data must be enabled; however, this can degrade performance when retrieving long data with Static cursors as the entire result set is stored on the client. To improve performance, you might consider designing your application to retrieve long data through forward-only cursors.

Encryption Method (EncryptionMethod): Data encryption may adversely affect performance because of the additional overhead (mainly CPU usage) required to encrypt and decrypt data.

Lock Timeout (LockTimeOut): Sometimes users attempt to select data that is locked by another user. Oracle provides three options when accessing locked data with SELECT ... FOR UPDATE statements:

Wait indefinitely for the lock to be released (-1)
Return an error immediately (0)
Return an error if the lock has not been released within a specific number of seconds (n seconds)

NOTE: This option is not supported by Oracle 8.

Some applications may benefit by not waiting indefinitely and continuing execution; this keeps the application from hanging. The application, however, needs to handle lock timeouts properly with an appropriate timeout value; otherwise, processing time could be wasted handling lock timeouts, and deadlocks could go undetected.

To improve performance, either enter a number of seconds or enter 0 as the value for this option.

Procedure Returns Results (ProcedureRetResults): The driver can be tuned for improved performance if your application's stored procedures do not return results. In this case, the Procedure Returns Results option should be disabled.

Server Process Type (ServerType): When using a dedicated server connection, a server process on UNIX (a thread on Windows) is created to serve only your application connection. When you disconnect, the process goes away. The socket connection is made directly between your application and this dedicated server process. This can provide tremendous performance improvements, but will use significantly more resources on UNIX servers. Because this is a thread on Oracle servers running on Windows platforms, the additional resource usage on the server is significantly less. This option should be set to 2 (dedicated) when you have a batch environment with lower numbers of connections, your Oracle server has excess processing capacity and memory available when at maximum load, or if you have a performance-sensitive application that would be degraded by sharing Oracle resources with other applications.

Use Current Schema for SQLProcedures (UseCurrentSchema): When this option is enabled, the driver returns only database objects owned by the current user when executing catalog functions. Calls to catalog functions are optimized by grouping queries. Enabling this option is equivalent to passing the Logon ID used on the connection as the SchemaName argument to the catalog functions.

Wire Protocol Mode (WireProtocolMode): Set this option to 2 if:

Your application executes Select statements that return more than one row, the rows returned have repeating data in some or all of the columns, and the repeated data is in consecutive rows (the data in column1/row1 is the same as the data in column1/row2, for example).
Your application updates or inserts images, pictures, or long text or binary data.

NOTE: If your application: 1) returns single row result sets or result sets that do not contain repeating data or 2) does not update or insert long data, this option should be set to 1; otherwise, performance may be degraded.