Pools

A pool is a specific collection of media to which Backup writes data. Backup uses pools to sort and store data. The configuration settings for each pool act as filters that tell Backup which volumes should receive specific data. Backup uses pools in conjunction with label templates to keep track of which data is on which specific volume. For detailed information about label templates, see "Labeling Storage Volumes ".

How Backup Uses Pools

The way you configure pools determines which volumes receive data. Each pool configuration contains a list of criteria that the data must meet for the data to be written to associated volumes. When you specify save sets to include in a pool, you can specify exact save set names, or you can use regular expression matching to send a group of save sets to a specific pool. For an example using regular expression matching, see "Example: Directing Client Indexes and Bootstrap to a Separate Pool ". For detailed information about regular expression matching, refer to the nsr_regexp man page.

When a scheduled backup occurs, Backup tries to match the save set to a pool configuration. If the save set matches the criteria of a pool configuration, Backup directs the save set to a labeled volume from that pool.

Backup then checks to see whether a correctly labeled volume is mounted on a storage device. If a correctly labeled volume is mounted on a storage device, Backup writes data to the volume. If a correctly labeled volume is not mounted on a storage device, Backup requests that such a volume be mounted and waits until an operator mounts the appropriate volume.

Backup Pool Types

Backup provides preconfigured pool types to keep different types of data separate. Backup does not mix the following types of data on volumes within a pool:

• Backup data

• Archive data

• Cloned data

• Migration data

Unless you specify other pools, all backup data is routed to the Default pool and all archive data is routed to the Archive pool. Cloned backup data is routed to the Default Clone pool, and cloned archive data is routed to the Archive Clone pool.

How Backup Uses Pool Criteria to Sort Data

When you configure Backup, you can create additional pools and sort data by pool type and any combination of the following criteria:

• Group (backup group)

• Backup client

• Save sets (file or filesystems)

• Backup levels (full, levels 1 - 9, incremental, manual)

If you begin by entering a group name in the Group attribute, the pool is immediately restricted to accept only data associated with the named group. If you add a second group name to the Group attribute, the pool accepts data associated with either group, but no others. Entries for a single attribute function as "OR" clauses; that is, the pool accepts data from clients in either group.

Each of the four configuration criteria, however, functions with the others as an "AND" clause. That is, if you enter configuration criteria in both the Group attribute and Save Set attribute, only data that meets both the Group criteria and the Save Set criteria is written to volumes from the specified pool.

You cannot create pools that share identical settings for pool type, group, client, save set, or level. If the settings for a new pool match the settings for an existing pool, you receive a warning message. Change the appropriate settings and reapply to save the pool resource.

For further information about save sets, see "Specifying Which Data Is Backed Up ". For further information about groups or backup levels, see "Backup Levels ".

Example: Directing Client Indexes and Bootstrap to a Separate Pool

You can use regular expression matching to direct the client indexes and bootstrap to a different pool than you send the backup data.

In the following example, the client file indexes are in /nsr/index. To send the Backup server's bootstrap and all the client file indexes from this filesystem to the same pool, create a pool (in the Pools resource) with the following attributes:

name: Index;
pool type: Backup;
save sets: bootstrap, /nsr/index/.*;
levels: ;

When the group's scheduled backup runs, the client save sets are written to a volume labeled for the appropriate save set pools, while the Backup server's bootstrap and /nsr/index save sets are written to a separate volume labeled for the "Index" pool.

When Data Meets the Criteria for More Than One Pool Configuration

Depending on the pool configurations you create, you might have data that matches the criteria for more than one pool configuration. For example, if you configure one pool to accept data from a group called "Accounting," and you configure another pool to accept data from all full backups, Backup has to determine to which pool a full backup for the Accounting group is written. Backup uses the following pool selection criteria:

1. Group (highest precedence)

2. Client

3. Save set

4. Level (lowest precedence)

When data matches the attributes for two pools, for example, Group and Level, the pool data is written to the pool specified in the Group attribute. In the example where the data matched the criteria for two pools, the one configured to accept data from the Accounting group and the other configured to accept data from all full backups, the data is routed to the pool that accepts data from the Accounting group.

When Data Does Not Meet the Criteria for Any Pool

When you use customized pool configurations to sort your data, you might inadvertently omit a client or save set. During a scheduled backup, if data does not meet the criteria for any customized pool configuration, Backup automatically sends the data to the Default pool. Backup uses the Default pool to ensure that all data for clients in a backup group is backed up to a volume.

When Backup sends data to the Default pool, Backup looks for a labeled volume from the Default pool mounted on a storage device. If no Default pool volume is mounted on a storage device, Backup requests the appropriate volume and waits until an operator mounts the volume. If Backup asks for a Default pool volume in the middle of a scheduled backup but an operator is not present to mount a Default pool volume, the backup pauses until an operator mounts a Default pool volume. If you have an operator available to monitor the backups, it is a good idea to keep a volume labeled for the Default pool close at hand in case this situation unexpectedly arises.

If you plan to use Backup for unattended backups, run a test of the backup after making any configuration changes to ensure that all data is written to the appropriate volumes and to avoid an unexpected Backup request for a Default pool volume. For the procedure to test your scheduled backup, see "Immediate Start of a Scheduled Group Backup".

Configuring a Pool for Incremental Backups

If you want to create a separate pool for incremental backups, be aware that the Backup hierarchy of precedence affects the way the data is stored. If the Level attribute value is "incremental," incremental data is routed to the associated pool but the corresponding changes to the client's file index are not. Backup saves all client file indexes at level 9 to speed the recovery operation, if one is needed.

If the client file indexes do not meet the criteria for the pool associated with the incremental backups, Backup matches the indexes to another pool (usually the Default pool) and looks for an appropriately labeled volume to write to. If you need to recover your data, you might have to use a large number of volumes to recover all your data. Thus, to store the client file indexes along with the incremental backup data and to speed the recovery operation, define the Level value in the Pools resource to accept both level 9 and incremental data.

You can use the Backup preconfigured NonFull pool settings to ensure that the client file indexes belong to the same pool as their incremental backups. When you keep the indexes in the same pool as their incremental backups, you reduce the number of volumes you need for a recovery.

Configuring a Pool for Clone Data

If you want to clone data, Backup requires a specific pool to receive the clone data, and a minimum of two devices, one to read the source volume and the other to write the clone. If you do not associate data to be cloned with a customized clone pool, Backup automatically uses the Default Clone pool. You must mount an appropriately labeled volume on a separate storage device for the cloning process to proceed smoothly. See "Cloning " for more information on the Backup cloning feature.

Configuring a Pool for Archive Data

If you want to use Backup Archive to archive data, Backup requires a specific pool to receive the archive data. You can then store these volumes offsite, if you want. If you do not associate data to be archived with a customized archive pool, Backup automatically uses the preconfigured Archive pool. You must mount an appropriately labeled volume on a storage device for the archive process to proceed smoothly. See Chapter 6, Backup Archive for more information on the Backup archive feature.

Configuring a Pool for Migration Data

If you use the HSM feature, Backup requires a specific pool to receive the premigrated and migrated save sets. If you do not associate the migration data with a customized migration pool, Backup automatically uses the preconfigured Migration pool. You must mount an appropriately labeled volume on a storage device for the premigration and migration processes to proceed smoothly. See Failed Cross Reference Format for more information on the Backup HSM feature.

Archive and migration data are in a different format than regular Backup save set data. Therefore, they must be written to different volumes. Because of these differences, the client file indexes and bootstrap save set created during an archive, premigration, or migration operation are also not written to the same volume as the archived or migrated save sets. By default, they are written to a volume from the Default pool. If you need to direct the client file indexes and bootstrap to a volume pool other than Default, see "Example: Directing Client Indexes and Bootstrap to a Separate Pool " for information.

Configuring a Pool for Manual Backups

You can create a customized pool to receive data from a manual backup by specifying "manual" in the Level attribute. Backup, however, sorts data from a manual backup differently than data from a regularly scheduled backup. Because a manual backup is not performed as part of a scheduled backup group, the data is not associated with any group name. Thus, when you perform a manual backup in which only a single client's save set data is saved, the group normally associated with that client's save set is not included as a criterion for pool assignment. As a consequence, data from a manual backup may be sent to a different pool than the pool in which data from this client's save set is stored during a regularly scheduled backup operation.

If you do not create a customized pool to receive data from manual backups, Backup uses the Default pool and looks for a mounted volume from the Default pool on which to write manually backed-up data. Because Backup tracks the volume location of all backup data, you do not need to worry about tracking which volume contains the manually backed-up data. If you need to recover the data, Backup requests the correct volume.

When you perform a manual backup, the client index and server bootstrap are not included in the backup. If you never perform regularly scheduled backups of the clients and server machines, the information vital to data recovery in the event of a disaster is not available.