7 ODBC Application Tuning

This chapter describes how to tune an ODBC application to run optimally on a TimesTen database. See "TimesTen Database Performance Tuning" in Oracle TimesTen In-Memory Database Operations Guide for more general tuning tips.

This chapter includes the following topics:

Bypass driver manager if appropriate

Applications that do not need ODBC access to database systems other than TimesTen should omit the driver manager. This is done by linking the application directly with the TimesTen direct or client driver, as described in "Linking options". The performance improvement is significant.

Using arrays of parameters for batch execution

You can improve performance by using groups, referred to as batches, of statement executions in your application.

The SQLParamOptions ODBC function enables an application to specify multiple values for the set of parameters assigned by SQLBindParameter. This is useful for processing the same SQL statement multiple times with various parameter values. For example, your application can specify multiple sets of values for the set of parameters associated with an INSERT statement, and then execute the INSERT statement once to perform all the insert operations.

TimesTen supports the use of SQLParamOptions with INSERT, UPDATE, DELETE, and MERGE statements, but not with SELECT statements.

TimesTen recommends the following batch sizes for TimesTen Release 18.1.

In TimesTen Classic:

  • 256 for INSERT statements

  • 32 for UPDATE statements

  • 32 for DELETE statements

  • 32 for MERGE statements

In TimesTen Scaleout:

  • 1024 x (number of elements in the grid) for INSERT statements

  • 32 x (number of elements in the grid) for UPDATE statements

  • 32 x (number of elements in the grid) for DELETE statements

(TimesTen Scaleout does not support MERGE statements.)

Table 7-1 provides a summary of SQLParamOptions arguments. Refer to ODBC API reference documentation for details.

Table 7-1 SQLParamOptions arguments

Argument Type Description

hstmt

SQLHSTMT

Statement handle

crow

SQLULEN

Number of values for each parameter

pirow

SQLULEN

Pointer to storage for the current row number

Assuming the crow value is greater than 1, the rgbValue argument of SQLBindParameter points to an array of parameter values and the pcbValue argument points to an array of lengths. (Also see "SQLBindParameter function".)

In the TimesTen Classic Quick Start, refer to source file bulkinsert.c for a complete working example of batching. (Also, for programming in C++ with TTClasses, see bulktest.cpp.) See "TimesTen Quick Start and sample applications".

Note:

When using SQLParamOptions with the TimesTen Client/Server driver, data-at-execution parameters are not supported. (An application can pass the value for a parameter either in the SQLBindParameter rgbValue buffer or with one or more calls to SQLPutData. Parameters whose data is passed with SQLPutData are known as data-at-execution parameters. These are commonly used to send data for SQL_LONGVARBINARY and SQL_LONGVARCHAR parameters and can be mixed with other parameters.)

Avoid excessive binds

The purpose of a SQLBindCol or SQLBindParameter call is to associate a type conversion and program buffer with a data column or parameter. For a given SQL statement, if the type conversion or memory buffer for a given data column or parameter is not going to change over repeated executions of the statement, it is better not to make repeated calls to SQLBindCol or SQLBindParameter. Simply prepare once and bind once to execute many times.

Avoid SQLGetData

SQLGetData can be used for fetching data without binding columns. This can sometimes have a negative impact on performance because applications have to issue a SQLGetData ODBC call for every column of every row that is fetched. In contrast, using bound columns requires only one ODBC call for each fetched column. Further, the TimesTen ODBC driver is more highly optimized for the bound columns method of fetching data.

SQLGetData can be very useful, though, for doing piecewise fetches of data from long character or binary columns. (This is discussed for LOBs in "Using the LOB piecewise data interface in ODBC".)

Avoid data type conversions

TimesTen instruction paths are so short that even small delays due to data conversion can cause a relatively large percentage increase in transaction time. To avoid data type conversions:

  • Match input argument types to expression types.

  • Match the types of output buffers to the types of the fetched values.

  • Match the connection character set to the database character set.

Bulk fetch rows of TimesTen data

TimesTen provides the TT_PREFETCH_COUNT ODBC statement option to enable an application to fetch multiple rows of data. This feature is available for applications that use the Read Committed isolation level. For applications that retrieve large amounts of TimesTen data, fetching multiple rows can increase performance greatly. However, locks are held on all rows being retrieved until the application has received all the data, decreasing concurrency. For more information on how to use TT_PREFETCH_COUNT, see "Prefetching multiple rows of data".

Optimize queries

TimesTen provides the TT_PREFETCH_CLOSE ODBC connection option to optimize query performance. For information on how to use this attribute, see "Optimizing query performance".