Database agents support the latest releases of all supported database engines this includes:
Support of JDBC 3.0 (JDK 1.4) is now available, with new JDBC Client drivers opljdbc3.jar & megathin3.jar for providing this connectivty. A new Sample program is also available for demonstrating the used of Cached RowSet in JDBC 3.0
Our Drivers are now built to communicate directly with Microsoft SQL Server and SYBASE ASE using the TDS protocol (the native wire protocol for both database servers). This also implies that no additional software is required post installation in order for our ODBC Drivers to communicate with these Drivers (this applies to the Single Tier format Drivers only). The use of TDS has also enabled us to double the performance of both our Single Tier and Multi-Tier Drivers for these databases.
All drivers now support the SQLSetStmtAttr(SQL_ARRAY_SIZE) call for batch select statements, providing improved performance when re-execute select statements with bound paramters.
The Release 4 OpenLink driver family brings marked communications layer improvements. Central to these are improved implementation of the SQLGetData and SQLPutData ODBC function calls.
In previous releases of the drivers, parameter data at query execution was assembled in the OpenLink driver's client component and transferred to its server component in a single network hop. (The client and server components apply to both the Single- and Multi-Tier drivers; they refer to different layers within the driver entity.) Similarly, when fetching from a "long", or large binary data column, data was transferred from the driver's server component to the client component in a single network transfer. The only way data could manipulated in a piecemeal fashion, was within the ODBC application from the driver's client component (client side only).
The Release 4 driver family now allow transferring parameter data in parts over the network between the client and server driver components. Once transferred, the fragmented column data are re-pieced together in their entirety within the client and server portions of the driver.
When fetching, data from columns with "long" data are only transferred between the server and client components if one of the following applies to that column:
This mechanism is referred to as deferred fetching. In this method, as "long" column data is not reassembled within the OpenLink ODBC client itself (rather, within the application), driver memory overhead incurred is dramatically reduced. Deferred fetching applies to the following "long" database column types:
| ODBC agent: |
SQL_LONGVARCHAR SQL_LONGVARBINARY |
| DB/2 agent: |
SQL_BLOB SQL_CLOB SQL_DBCLOB If the long data compatibility option has been specified in the DB2 database SQL_LONGVARCHAR SQL_LONGVARBINARY SQL_LONGVARGRAPHIC |
| Oracle agent: |
SQLT_BLOB SQLT_CLOB |
| Sybase agent: |
CS_IMAGE_TYPE CS_TEXT_TYPE |
However, if a table contains a column defined as one of these "long" types, but the actual data stored in the column only fills a small proportion of the available space, deferred fetching is of no benefit. In these cases, performance may be improved by switching off the deferred fetching mechanism using the control in the OpenLink Generic Client data source setup dialog.
|
Previous
JDBC 3 Driver Classes |
Chapter Contents |
Next
Contents of Distributed Transaction Processing (XA) |
