Threading

The JDBC driver is completely thread safe; that is, it will not fail when database requests are made on separate threads.

Threading Architecture

A JDBC driver can be based on one of the following architectures:

Thread impaired. The JDBC driver serializes all JDBC calls. All requests are handled one by one, without concurrent processing.
Thread per connection. The JDBC driver processes requests concurrently with statements that do not share the same connection; however requests on the same connection are serialized. The SequeLink for JDBC driver uses this architecture.
Fully threaded. All requests use the threaded model. The JDBC driver processes all requests on multiple statements concurrently.

Cancelling Functions in Multithreaded Applications

In a multithreaded application, a thread can use the cancel method to cancel a statement that is being executed by another thread. Whether the cancel method actually cancels the statement depends on the data store being accessed as shown in Table 8-2.

OK means that cancel can interrupt the running statement.
Ignored means that cancel will have no affect on the running statement.

In both cases, the cancel method returns SQL_SUCCESS. If the cancel method has been called from a different thread while a request is pending, the original statement will return SQL_ERROR with the error message Operation cancelled.

Table 8-2. Using Cancel in Multithreaded JDBC Applications 

Data Store	SQLCancel
DB2 UDB on z/OS	Ignored
DB2 UDB on Windows	OK
DB2 UDB on Linux/UNIX	Ignored
Informix	OK
Microsoft SQL Server	OK
Oracle on Windows	Ignored
Oracle on UNIX	OK
Sybase	OK
NOTE: Cancel functionality is not supported when the connection uses Secure Socket Layer (SSL) encryption.