How commitment control works with objects

When you place an object under commitment control, it becomes a committable resource. It is registered with the commitment definition. It participates in each commit operation and rollback operation that occurs for that commitment definition.

The following topics describe these attributes of a committable resource:

• Resource type
• Location
• Commit protocol
• Access intent

• Types of committable resources
  The table in this topic lists the different types of committable resources.
• Local and remote committable resources
  A committable resource can be either a local resource or a remote resource.
• Access intent of a committable resource
  The access intent determines how the resources participate together in a transaction.
• The commit protocol of a committable resource
  Commit protocol is the capability a resource has to participate in one-phase or two-phase commit processing. Local resources, except API committable resources, are always two-phase resources.
• Journaled files and commitment control
  You must journal (log) a database file (resource type FILE or DDM) before it can be opened for output under commitment control or referenced by an SQL application that uses an isolation level other than No Commit. A file does not need to be journaled in order to open it for input only under commitment control.
• Sequence of journal entries under commitment control
  This table shows the sequence of entries that are typically written while a commitment definition is active. You can use the Journal entry information finder to get more information about the contents of the journal entries.
• Commit cycle identifier
  A commit cycle is the time from one commitment boundary to the next. The system assigns a commit cycle identifier to associate all of the journal entries for a particular commit cycle together. Each journal that participates in a transaction has its own commit cycle and its own commit cycle identifier.
• Record locking
  When a job holds a record lock and another job attempts to retrieve that record for update, the requesting job waits and is removed from active processing.