IBM i series Handbook page 501

controllers and placement of these controllers on separate buses can give even higher
protection.
• Journaling: Provides the ability to record all changes to records in a file as they occur.
These journaled changes are applied to the file if the system is lost. With the
implementation of remote journals, this protection can be extended to cover access
paths to provide faster recovery of access paths in the event of an abnormal system
termination.
A single application may access files logged to multiple journals. If the transaction adjusts
data in several journal receivers, two-phase commitment control is employed to
synchronize the information located in all receivers. If only files from a single receiver are
used during the execution of the transaction, the commitment cycle focuses only on that
journal with a single-phase commitment.
In V5R1, journaling has been enhanced for high availability, with the addition of byte
stream file and the ability for directories to be journaled. This allows a customer to use
third-party software to provide replication on an other iSeries server or to use the
journaled information for other recovery and monitoring purposes.
• Commitment control: Ensures that if a transaction requires multiple database
changes, all of them (or none of them) are made. A single transaction may involve
several database changes. Use of commitment control allows the application program
to treat all changes for a transaction as a single group – to be either committed in the
event of normal completion or backed out in the event of abnormal completion. An
application programmer can easily provide for recovery at a transaction boundary
using HLL statement to request commitment or rollback. Commitment control requires
journaling.
• Save-While-Active (SWA): Allows one or more libraries to be saved while operations,
including changes, continue against the libraries. During a short period of quiesced
operation, a checkpoint is taken of all libraries being saved before the first save
operation begins, so that all libraries are synchronized.
• RAID-5 disk protection: Can be implemented with the use of a disk controller and at
least four disks to make up an array.
• Save/restore to multiple tapes concurrently- Parallel Save/Restore: The SAVLIB,
RSTLIB, SAVOBJ, RSTOBJ, and SAVCHGOBJ commands and the QSRSAVO API
support using multiple tape devices, or multiple resources in a tape library, in parallel.
This support reduces the amount of time required to save and restore very large
objects. Previously, the maximum save rate for large objects was limited to the
maximum throughput capabilities of a single tape device. By using multiple tapes in
parallel, the maximum throughput can be increased and time-to-save is reduced,
especially for very large objects. Recovery from this set of parallel save tapes is most
efficient when the same number of tape devices is used on the restore as on the save
(from two to 32).
Operating System/400 (5722-SS1)
Operating System/400 (5722-SS1)
471