Performance improvements for distributed applications
Db2 11 provides improved performance for distributed applications that return large result sets.
Db2 11 introduces package-based continuous block fetch, which can improve performance for retrieval of large, read-only result sets from a remote Db2 for z/OS® server.
Like SQL-based continuous block fetch, package-based continuous block fetch causes fewer messages to be transmitted from the requester to retrieve the entire result set. However, package-based continuous block fetch is easier to configure. It requires only that you bind your applications with the new DBPROTOCOL(DRDACBF) option. You do not need to modify your applications or set subsystem parameters to indicate the maximum number of blocks to be returned for a remote request.
In addition, package-based continuous block fetch is more efficient than SQL-based continuous block fetch. With package-based continuous block fetch, the requester opens a secondary connection to the Db2 server for each read-only cursor. The Db2 server returns extra query blocks until all of the rows for the cursor are retrieved. When the cursor is closed, the secondary connection is implicitly closed.
Package-based continuous block fetch provides a performance advantage for a Db2 for z/OS application that has the following characteristics:
- The application queries only remote sites.
- The application does not contain INSERT, UPDATE, DELETE, or MERGE statements.
- No statement in the application creates a unit of recovery on the remote site. This situation results in an SQL error when the application package is bound for package-based continuous block fetch.