Each scheduler requires a database in which to store its persistent information. Schedulers use this database for storing tasks and then running them. The choice of database and location should be determined by the application developer and server administrator.
Scheduler performance is ultimately limited by database performance. If you need more tasks per second, you can run the scheduler daemons on larger systems, use clusters for the session beans used by the tasks or partition the tasks by using multiple schedulers. Eventually, however, the scheduler database becomes saturated, and a larger or better-tuned database system is needed. For detailed information on scheduler topologies refer to the "WebSphere® Enterprise Scheduler planning and administration guide" technical paper.
Multiple
schedulers can share a database when you specify unique table prefix
values in each scheduler configuration. This sharing can lower the
cost of administering scheduler databases. However,
do not configure schedulers with non-unique table prefixes such that
two separate servers share the same database table. A lease occurs
between a specific database table and a scheduler running on a server.
This lease allows one server at a time to own the lease to a specific
database table. This process exists to ensure that one server runs
schedule events, such as Enterprise JavaBeans (EJB) timers, in a cluster
environment. If the server with this lease is unavailable, another
server in the cluster obtains the lease.
In this information ...Subtopics
Related concepts
| IBM Redbooks, demos, education, and more(Index) |