Introduction to FDRINSTANT and FlashCopy
FDRINSTANT and FlashCopy
FDRINSTANT™ is an additional cost FDR facility for non-disruptive backup. It works in concert with FlashCopy®, a feature of many IBM DASD supported by the major DASD subsystems that allows for the “instant” duplication of existing DASD volumes and data sets.
FDRINSTANT enhances:
- FDR full volume backup
- FDRDSF data set backup
- FDRCOPY data set copy
- FDRABR volume backup (full-volume and incremental backup).
It provides non-disruptive backup of offline operating system DASD volumes, both SMS-managed and non SMS-managed, using technology developed by BMC.
FDRINSTANT supports “Space Efficient” (SE) FlashCopy. FDRINSTANT automatically detects a Space Efficient flash unit designation and appropriately establishes the FlashCopy relationship.
Traditional backups may be disruptive of normal operations. It obviously takes time to backup a data set or volume. If data sets are being updated by some application while they are being backed up, the backup may not be valid and the data sets may not be usable when they are restored. The usual response to this problem is to quiesce all updates to the data during the backup, disrupting normal operation until the backup is complete.
When used with FlashCopy, FDRINSTANT allows you to instantly create a point-in-time copy of an online source volume. This point-in-time copy can be used for backup processing while the source volume is available for updates. Normal FDR volume and data set serialization techniques ensure the integrity of the point-in-time image. FlashCopy also allows for the rapid creation of copies of data sets.
FDRINSTANT enhances FDR components
FDRINSTANT enhances FDR, FDRDSF, FDRCOPY, and FDRABR to utilize the “point-in-time” image created by FlashCopy.
FDRINSTANT enhances FDRCOPY with the fast replication capabilities of FlashCopy. Data sets can be copied or moved almost instantly within the same DASD subsystem. FDRINSTANT provides this fast replication capability transparently without control card modifications.
If you are licensed for FDRABR, FDRINSTANT enhances ABR Volume Backups, both full-volume (TYPE=FDR) and incremental (TYPE=ABR/DSF/AUTO) to use FlashCopy, by creating “instant” point-in-time copies of the volumes to be backed up, and then moving that captured data to tape.
FlashCopy
FlashCopy is an additional cost feature of DASD subsystems that provides a means to create a “point-in-time” copy or image of a DASD volume or extents.
When FlashCopy is invoked, a FlashCopy session is started between the two DASD volumes or two data sets involved. If the session is started with the FlashCopy COPY option, the subsystem copies all specified tracks to the target volume. FlashCopy also supports a NOCOPY option, where the tracks are not physically copied unless necessary.
The NOCOPY option of FlashCopy is used when the copied data is only needed for a short amount of time to perform some task. Only tracks being updated are copied to the target to retain the point-in-time image. Once the task completes, the FlashCopy relationship is withdrawn and that target volume is no longer valid. NOCOPY is used by FDRINSTANT in its dump processing.
The COPY option of FlashCopy is used when the target volume is to retain the data indefinitely. All of the data tracks are copied to the target. FCOPY=COPY is the default in FDRINSTANT FDRCOPY processing.
When an application reads a track from the FlashCopy target that has not yet been physically copied from the source, the subsystem actually reads that track from the source. If a track that has been copied to the target is requested, it is read from the target. Therefore, when the target volume or data set is read, the application reads the tracks exactly as they existed on the source at the beginning of the FlashCopy session.
FlashCopy without FDRINSTANT
Although FlashCopy can create a point-in-time image of a production DASD volume, traditional backup products cannot directly use the copied image without additional considerations.
For example, remember that a full-volume FlashCopy copies all of the original volume’s data tracks, including the label track containing the volume serial. Since the operating system does not let you put two volumes with the same volume serial online, the new volume is forced offline. Various backup products require that the duplicated volume be brought online, so the volume serial of the target volume must be changed.
It is possible to preserve the original volume serial of the target volume or use ICKDSF to re-label the target volume after the FlashCopy session is started, so the volume can be put online. This invokes additional requirements for SMS-managed volumes, VSAM clusters, and cataloged data sets. The data sets on the source volume appear to be uncataloged, which may cause problems during backup and restore unless they are renamed and re-cataloged.
A backup of a re-labeled volume appears to be a backup from the new volume serial, not the original volume serial, so additional procedures are required to document this correspondence (for example, backup of volume “B” is really a backup of volume “A”). Restore procedures must be modified to restore the backups to the correct serial (for example, restore volume “B” back to volume “A” and re-label it as “A”).
The FDRINSTANT Solution
FDRINSTANT allows you to create a duplicate volume image with FlashCopy without changing its volume serial or bringing it online. FDR full volume backup, FDRDSF data set backup, FDRCOPY data set copy, and FDRABR volume backups use FDRINSTANT technology to read the duplicated volume even while it remains offline. A non-disruptive backup or copy can start as soon as the FlashCopy session is started. Updates to the original volumes can continue while you create the point-in-time backup or copy of the data.
For FDRABR, FDRINSTANT makes volume backups a two-phase process. The first phase (“Instant Phase”) executes an FCOPY statement under FDRABR, to start a FlashCopy session for each selected volume to an offline volume in the same subsystem at an address that you provide. This takes only seconds per volume. The copied volume is actually a valid backup at that point-in-time; you can do a restore from it if full-volume or data set recovery is required. The second phase (“Dump Phase”) executes a normal FDRABR full-volume or incremental backup of the same volumes with a parameter that tells ABR to read the offline point-in-time image and move it to a backup tape. There is no need to identify the target device to the ABR backup step; it knows the address of the most recent copy of each online volume.
FDRINSTANT also makes FDR and FDRDSF backups a two-phase process. The first phase (“Instant Phase”) executes a FCOPY statement under program FDRFLASH, to start a FlashCopy session for each selected volume to an offline volume in the same logical subsystem at an address that you provide. When that step ends, the data has been logically copied and you can resume normal operations on the original volume. Then you execute an FDR or FDRDSF backup with a parameter that tells FDR to read the copy instead of the original volume. It then reads the offline copy in exactly the same way that it would read the online volume. FDRINSTANT produces a backup that appears to be a conventional backup of the original volume serial. FDRINSTANT blends its unique off-line, high-speed non-disruptive backup together with a traditional restore complete with FDR’s powerful logical file capability.
There are no special considerations for restores from a backup created by FDRINSTANT. You can restore entire volumes or individual data sets from those backups. The target volumes for restore are normal online volumes. Since the backups created by FDRINSTANT appear to be backups of the original volume, there is no special volume serial or data set name concerns.
FDRCOPY data set copy and move can be used with FDRINSTANT in two ways:
- When FDRCOPY detects that the input and output volume associated with a given data set are in the same subsystem with the FlashCopy feature installed, it automatically invokes FlashCopy to copy the tracks, resulting in an almost instantaneous copy. Normal read/write is used if the volumes are not both in the same subsystem or if the feature is not installed. If you want to force the use of normal read/write, specify the operand FCOPY=NO on the COPY or MOVE statement.
- Alternatively, you can use FlashCopy to freeze an image of the volumes containing the desired data sets, and then use FDRCOPY to create copies of those data sets on other online volumes. This is useful if the target volumes are not in the same subsystem as the source volumes, allowing you to capture an instant image of the source data sets and copy them at your leisure.
