Data Facility Storage Management Subsystem full volume dump
Instead of using Hierarchical Storage Manager (HSM), you can use DFSMSdss directly to dump and restore full volumes to and from cloud object storage by using Transparent Cloud Tiering (TCT) with the full volume dump (FVD) function. These full volume backups can then be used to repair or recover a production environment that was corrupted by either a system failure, a human error, or compromised by either a cyberattack or internal fraud.
This chapter describes how to implement FVD to cloud object storage, which preparation steps you need to consider, how you can manage (list and delete) FVD backups in the cloud and for which use cases the FVD is appropriate. Finally, it explains how to restore an FVD from the cloud target.
|
Note: DFSMSdss does not support restoring individual data sets from FVDs. If you need to restore individual data sets, you can restore the volume, use ICKDSF to condition the volume, bring it online and then copy the data sets by using physical data set COPY.
|
The data flow and the way FVDs are stored in the cloud are the same as described in Chapter 3, “Transparent Cloud Tiering” on page 19, where you can also find the supported cloud targets.
How to configure those cloud targets including the IBM TS7700 is described in Chapter 6, “Configuring the IBM DS8000 for Transparent Cloud Tiering” on page 49.
This chapter includes the following topics:
13.1 FVD for TCT overview
FVD enables you to create not just backups of inactive but also active data to be stored in a cloud target. The FVD data flow is directly between the DS8000 and the target cloud, and does therefore not consume mainframe central processing unit (CPU) cycles.
You can create offline copies of your data, which can be used to protect your business against data loss and data corruption, thus meeting regulatory and compliance guidance. In any malicious event affecting your data, you can perform an FVD restore of your stored data in the cloud target back into your production environment.
The FVD to the cloud is in principle a standard DFSMSdss FVD whose target is an object store. The difference and specifics of the FVD to cloud are:
1. The configuration of the cloud target as described in Chapter 6, “Configuring the IBM DS8000 for Transparent Cloud Tiering” on page 49.
2. The implementation of your cloud credentials handling.
3. The new parameters to be used in the FVD job to target the cloud.
4. The handling and managing of the dumps with new DFSMSdss commands.
|
Note: When accessing the cloud object storage, Data Facility Storage Management Subsystem (DFSMS) needs the cloud credentials (user ID and password). To avoid having to specify the password in clear text in all job definitions, use the CDACREDS method to store the credentials securely, as described in 8.1, “Managing credentials by using Cloud Data Access” on page 80.
|
13.1.1 FVD use cases and how it works
You can implement FVD to cover the following use cases:
•Create point-in-time (PiT) backups of IBM Db2 image copies by using DFSMSdss and FlashCopy.
•Create PiT volume copies by using DFSMSdss and FlashCopy with Dump Conditioning.
•Retain versions of Safeguarded Copy backups.
The FVD may be performed against FlashCopy target volumes, or directly from production volumes. Dumping from a FlashCopy target is a best practice because it reduces the amount of time that the production volume is reserved. Using dump conditioning, this volume dump looks as if it was taken directly from the production volumes and not from the intermediate FlashCopy volumes. You can use the same methods to create these FlashCopy copies as you would use for classical dumps to disk or tape.
If you use aTS7700 as object storage target, you can enable compression and save significant space in the TS7700 object storage partition. For more information, see 3.9, “Transparent Cloud Tiering compression with TS7700” on page 32.
The workflow, which performs an FVD from the FlashCopy volumes targeting to the cloud is illustrated in Figure 13-1 on page 127. The workflow is the same when you are running an FVD of Db2 image copies.
Figure 13-1 Performing an FVD to a cloud object store workflow
Another use case, which performs an FVD from Safeguarded Copy recovery volumes targeting the cloud is illustrated in Figure 13-2. The number of Safeguarded Copy copies that you can keep online is finite, and if you must keep more backups than you can keep with Safeguarded Copy, you can periodically recover a volume and dump it to the cloud.
Figure 13-2 FVD from a Safeguarded Copy recovery volume
Whether you are restoring from a backup of a production volume, a FlashCopy target or a Safeguarded Copy recovery volume, there is no need to manually recall the data before the restore can be done. The restore is directly performed to the production volumes. If the data was compressed, it is decompressed before it is written to the production volumes.
The corresponding FVD recovery workflow is depicted in Figure 13-3.
Figure 13-3 FVD restore workflow
13.2 Creating an FVD to cloud object storage
The job that you establish to run an FVD looks like a usual FVD with the following new parameters:
•CLOUD(cloudname): Invoke the cloud target that was defined in DFSMS.
•CONTAINER(containername): Targets the specified container that is defined in the cloud object storage. If the container does not yet exist, it is created when running the FVD.
•CDACREDS: Tells the FVD to use the stored CDA credentials when connecting to the object storage.
•OBJECTPREFIX('objectprefix’): This prefix makes the created FVD definite so that it can recognize FVDs that were created for the same data. A best practice approach for the prefix is to use IBM z/OS variables if you plan to perform regular FVDs of the same data. Some examples are:
– &DATE/&TIME
– &JOBNAME/&PREFIX
For more information, see Coding variables in JCL.
The JCL snippet in Example 13-1 on page 129 shows a sample dump job. In this example, the cloud name is TS7700CLOUD that targets an IBM TS7700 object partition. The container name is defined as DSSDUMPS. We use the predefined CDA credentials to access the cloud. The object prefix is set to the jobname of the FVD job followed by the VOLSER for which the FVD is performed and the date when the FVD was created.
The SYSPRINT in this example is written into a IBM z/OS Generation Data Group (GDG) file, which is a best practice to track the created FVD information.
Example 13-1 Sample job definition to create an FVD
//STEP0001 EXEC PGM=ADRDSSU
//SYSPRINT DD DISP=(NEW,CATLG),DSN=MY.CLOUD.INVENTORY(+1)...
//DISK1 DD UNIT=SYSDA,DISP=SHR,VOL=SER=VOL001
//DISK2 DD UNIT=SYSDA,DISP=SHR,VOL=SER=VOL002
//DISK3 DD UNIT=SYSDA,DISP=SHR,VOL=SER=VOL003
//SYSIN DD *,SYMBOLS=(EXECSYS)
PARALLEL
DUMP INDDNAME(DISK1) CLOUD(TS7700CLOUD) -
CONTAINER(DSSDUMPS) CDACREDS -
OBJECTPREFIX('&JOBNAME./VOL001/&DATE.’)-
WAIT(0,0) ALLDATA(*) ALLEXCP
DUMP INDDNAME(DISK2) CLOUD(TS7700CLOUD) -
CONTAINER(DSSDUMPS) CDACREDS -
OBJECTPREFIX('&JOBNAME./VOL002/&DATE.’)-
WAIT(0,0) ALLDATA(*) ALLEXCP
DUMP INDDNAME(DISK3) CLOUD(TS7700CLOUD) -
CONTAINER(DSSDUMPS) CDACREDS -
OBJECTPREFIX('&JOBNAME./VOL003/&DATE.’)-
WAIT(0,0) ALLDATA(*) ALLEXCP
For more information and a complete list of the options, see DUMP FULL command syntax, found at:
13.3 Restoring an FVD from cloud object storage
A restore of an FVD in the cloud is like a restore from a classical dump on disk or tape, except the additional parameters you need to define the cloud object specifics. In Example 13-2 on page 129, we provide a sample job definition for a restore from the cloud.
Example 13-2 Sample job definition to restore an FVD 
//STEP0002 EXEC PGM=ADRDSSU
//DISK1 DD UNIT=SYSDA,DISP=SHR,VOL=SER=VOL001
//SYSPRINT DD SYSOUT=A
//SYSIN DD *
RESTORE -
FULL COPYVOLID -
OUTDDNAME (DISK1) -
WAIT(0,0) PURGE -
CLOUD(TS7700CLOUD) -
CDACREDS -
CONTAINER(DSSDUMPS) -
OBJECTPREFIX('DNIGHTLY/VOL001/20201028')
The restore is performed directly to the production volumes as shown in Figure 13-3 on page 128. The example JCL will restore the FVD of volser VOL001 created the 2020-10-28 with the job named DNIGHTLY.
For more information about the DFSMSdss restore syntax, see RESTORE FULL command syntax.
13.4 Managing FVD backups
Along with the ability to do FVD to cloud, DFSMSdss provides the capability to manage those dumps. Other than IBM DFSMShsm migrated data, DFSMSdss does not manage dumps automatically. There is no automatic expiry and deletion, and no space management. You must track your dumps yourself and delete the one that you do not need anymore.
For this purpose, DFSMS supports the CLOUDUTILS LIST and CLOUDUTILS DELETE commands. They manage your FVD backups to list and delete them as required.
– Creates a list of all the containers in a cloud, all the dumps within a container or a subset of dumps in a container depending on how you specify the parameters CONTAINER and OBJECTPREFIX:
• Specifying CONTAINER lists all available FVDs in that container.
• With the OBJECTPREFIX parameter, it shows all the FVDs in the specified container with the defined prefix.
• With no CONTAINER and no OBJECTPREFIX specified, the command lists all available containers.
• With no OBJECTPREFIX specified, the command lists all available dumps in a container.
• OBJECTPREFIX supports the '*' wildcard.
– Access to the CLOUDUTILS LIST command requires at least read access to STGADMIN.ADR.CLOUDUTILS facility class profile.
Example 13-3 Cloud list creation JCL
/*
//STEP0001 EXEC PGM=ADRDSSU
//SYSPRINT DD SYSOUT=*
//SYSIN DD *
CLOUDUTILS LIST CLOUD(TS7700CLOUD) CDACREDS -
CONTAINER(DSSDUMPS) -
OBJECTPREFIX('*')
//
The command that is issued in the example lists all available FVD for the object store TS7700CLOUD in the specified container DSSDUMP, as no specific prefix is defined but a wildcard.
– Deletes one ore more dumps in a container or entire containers.
– CLOUDUTILS DELETE considerations:
• It is required to specify a CONTAINER.
• OBJECTPREFIX supports the '*' wildcard, and it is optional.
• Use FORCE to delete an entire non-empty container.
– Access to the CLOUDUTILS DELETE command requires at least read access to STGADMIN.ADR.CLOUDUTILS.DELETE facility class profile.
– Deleting non-empty containers with CLOUDUTILS DELETE FORCE requires at least read access to the STGADMIN.ADR.CLOUDUTILS.FORCE facility class profile.
Example 13-4 Deleting FVD backups in the cloud JCL
/*
//STEP0001 EXEC PGM=ADRDSSU
//SYSPRINT DD SYSOUT=*
//SYSIN DD *
CLOUDUTILS DELETE CLOUD(TS7700CLOUD) CDACREDS -
CONTAINER(DSSDUMPS) -
OBJECTPREFIX('DNIGHTLY/VOL001/202001*')
//
The JCL deletes all FVDs that are created with the prefix DNIGHTLY/VOL001/202001* in the container DSSDUMPS, which in this case is a Storage Group on the IBM TS7700 object store partition.
13.4.1 Use case example
Suppose a client creates nightly FVDs of a set of volumes and keeps one week worth of dumps. Within the SYSZADR.DSSDUMP container, we have a week worth of backups for vol001 and vol002. Figure 13-4 shows an example of a SYSZADR content for the container DSSDUMP.
Figure 13-4 Content of SYSZADR.DSSDUMP
We create another set of dumps with the prefix ‘vol00x/20203’, and if those dumps are successful, we delete all of the previous day’s created dumps for ‘vol00x/20202’.
For that purpose, we create a new version of the FVD. and if the step is successful, we delete the previous version, as shown in Example 13-5 on page 132.
Example 13-5 Creating a new version of the specific FVD and deleting the previous version
//STEP0001 EXEC PGM=ADRDSSU
//SYSPRINT DD DISP=(NEW,CATLG),DSN=MY.CLOUD.INVENTORY(+1)
//DISK1 DD UNIT=SYSDA,DISP=SHR,VOL=SER=VOL001
//DISK2 DD UNIT=SYSDA,DISP=SHR,VOL=SER=VOL002
//SYSIN DD *,SYMBOLS=(EXECSYS)
PARALLEL
DUMP INDDNAME(DISK1) CLOUD(TS7700CLOUD) -
CONTAINER(DSSDUMPS) CDACREDS -
OBJECTPREFIX('&JOBNAME./VOL001/20203.’)-
WAIT(0,0) ALLDATA(*) ALLEXCP
DUMP INDDNAME(DISK2) CLOUD(TS7700CLOUD) -
CONTAINER(DSSDUMPS) CDACREDS -
OBJECTPREFIX('&JOBNAME./VOL002/20203.’)-
WAIT(0,0) ALLDATA(*) ALLEXCP
SERIAL
IF MAXCC=0 THEN
DO
CLOUDUTILS DELETE CLOUD(TS7700CLOUD) –
CONTAINER(DSSDUMPS) CDACREDS -
OBJECTPREFIX('&JOBNAME./VOL001/20202’)-
CLOUDUTILS DELETE CLOUD(TS7700CLOUD) –
CONTAINER(DSSDUMPS) CDACREDS -
OBJECTPREFIX('&JOBNAME./VOL002/20202’)-
END
This use case is one specific use case that can be adapted to meet your requirements and it can be improved and automated by implementing a script that prepares the JCL needed. You can cover different cases when to delete previous created FVD, depending on your policy on how long you want to keep FVD backups available
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.