Tivoli Header

Administrator's Guide


Reclaiming Space in Sequential Access Storage Pools

Space on a sequential volume becomes reclaimable as files expire or are deleted from the volume. For example, files become obsolete because of aging or limits on the number of versions of a file. In reclamation processing, the server rewrites files on the volume being reclaimed to other volumes in the storage pool, making the reclaimed volume available for reuse.

The server reclaims the space in storage pools based on a reclamation threshold that you can set for each sequential access storage pool. When the percentage of space that can be reclaimed on a volume rises above the reclamation threshold, the server reclaims the volume. See the following sections:

How Tivoli Storage Manager Reclamation Works

Choosing a Reclamation Threshold

Reclaiming Volumes in a Storage Pool with One Drive

Reclamation for WORM Optical Media

Reclamation of Volumes with the Device Type of SERVER

Reclamation for Copy Storage Pools

How Collocation Affects Reclamation

Note:
Storage pools using the NETAPPDUMP data format are unable to use reclamation.

How Tivoli Storage Manager Reclamation Works

When the percentage of reclaimable space on a volume exceeds the reclamation threshold set for the storage pool, the volume is eligible for reclamation. The server checks whether reclamation is needed at least once per hour and begins space reclamation for eligible volumes. You can set a reclamation threshold for each sequential access storage pool when you define or update the pool.

During space reclamation, the server copies files that remain on eligible volumes to other volumes. For example, Figure 23 shows that the server consolidates the files from tapes 1, 2, and 3 on tape 4. During reclamation, the server copies the files to volumes in the same storage pool unless you have specified a reclamation storage pool. Use a reclamation storage pool to allow automatic reclamation for a storage pool with only one drive.

The server also reclaims space within an aggregate file. An aggregate is a physical file that contains multiple logical files that are backed up or archived from a client in a single transaction. Space within the file becomes reclaimable space as logical files in the aggregate expire or are deleted by the client. The server removes unused space from expired or deleted logical files as the server copies the aggregate file to another volume during reclamation processing. However, reclamation does not aggregate files that were originally stored in non-aggregated form. Reclamation also does not combine aggregates to make new aggregates. You can also reclaim space in an aggregate by issuing the MOVE DATA command. See Reclaiming Space in Aggregates During Data Movement for details.

Figure 23. Tape Reclamation

Tape Reclamation

After the server moves all readable files to other volumes, one of the following occurs for the reclaimed volume:

Volumes that have a device type of SERVER are reclaimed in the same way as other sequential access volumes. However, because the volumes are actually data stored in the storage of another TSM server, the reclamation process can consume network resources. See Reclamation of Volumes with the Device Type of SERVER for details of how the server reclaims these types of volumes.

Volumes in a copy storage pool are reclaimed in the same manner as a primary storage pool except for the following:

See Reclamation for Copy Storage Pools for details.

Choosing a Reclamation Threshold

The reclamation threshold indicates how much reclaimable space a volume must have before the server reclaims the volume. Space is reclaimable because it is occupied by files that have been expired or deleted from the TSM database, or because the space has never been used.

The server checks whether reclamation is needed at least once per hour. The lower the reclamation threshold, the more frequently the server tries to reclaim space. Frequent reclamation optimizes the use of a sequential access storage pool's space, but can interfere with other processes, such as backups from clients.

If the reclamation threshold is high, reclamation occurs less frequently. A high reclamation threshold is useful if mounting a volume is a manual operation and the operations staff is at a minimum.

Each reclamation process requires at least two simultaneous volume mounts, that is, at least two mount points (drives). The two drives must be in the same device class to allow the server to move the data from reclaimed volumes to other volumes in the same storage pool. A sufficient number of volumes, drives (if appropriate), and mount operators (if appropriate) must be available to handle frequent reclamation requests. For more information about mount limit, see Mount Limit. If the device class for the storage pool does not have two drives, you can specify a reclamation storage pool. For information about how to use a reclamation storage pool for storage pools with only one mount point, see Reclaiming Volumes in a Storage Pool with One Drive.

If you set the reclamation threshold to 50% or greater, the server can combine the usable files from two or more volumes onto a single new volume.

Setting the reclamation threshold to 100% prevents reclamation from occurring. You might want to do this to control when reclamation occurs, to prevent interfering with other server processes. When it is convenient for you and your users, you can lower the reclamation threshold to cause reclamation to begin.

Lowering the Migration Threshold

If you have been running with a high migration threshold and decide you now need to reclaim volumes, you may want to lower the threshold in several steps. Lowering the threshold in steps ensures that volumes with the most reclaimable space are reclaimed first. For example, if you had set the high migration threshold to 100%, first lower the threshold to 98%. Volumes that have reclaimable space of 98% or greater are reclaimed by the server. Lower the threshold again to reclaim more volumes.

If you lower the reclamation threshold while a reclamation process is active, the reclamation process does not immediately stop. If an onsite volume is being reclaimed, the server uses the new threshold setting when the process begins to reclaim the next volume. If offsite volumes are being reclaimed, the server does not use the new threshold setting during the process that is running (because all eligible offsite volumes are reclaimed at the same time).

Use the CANCEL PROCESS command to stop a reclamation process.

Reclaiming Volumes in a Storage Pool with One Drive

When a storage pool has only one mount point (that is, just one drive) available to it through the device class, data cannot be reclaimed from one volume to another within that same storage pool.

To enable volume reclamation for a storage pool that has only one mount point, you can define a reclamation storage pool for the server to use when reclaiming volumes. When the server reclaims volumes, the server moves the data from volumes in the original storage pool to volumes in the reclamation storage pool. The server always uses the reclamation storage pool when one is defined, even when the mount limit is greater than one.

If the reclamation storage pool does not have enough space to hold all of the data being reclaimed, the server moves as much of the data as possible into the reclamation storage pool. Any data that could not be moved to volumes in the reclamation storage pool still remains on volumes in the original storage pool.

The pool identified as the reclamation storage pool must be a primary sequential storage pool. The primary purpose of the reclamation storage pool is for temporary storage of reclaimed data. To ensure that data moved to the reclamation storage pool eventually moves back into the original storage pool, specify the original storage pool as the next pool in the storage hierarchy for the reclamation storage pool. For example, if you have a tape library with one drive, you can define a storage pool to be used for reclamation using a device class with a device type of FILE:

define stgpool reclaimpool fileclass maxscratch=100

Define the storage pool for the tape drive as follows:

define stgpool tapepool1 tapeclass maxscratch=100
reclaimstgpool=reclaimpool

Finally, update the reclamation storage pool so that data migrates back to the tape storage pool:

update stgpool reclaimpool nextstgpool=tapepool1

Reclamation for WORM Optical Media

Reclamation for WORM volumes does not mean that you can reuse this write-once media. However, reclamation for WORM volumes does allow you to free library space. Reclamation consolidates data from almost empty volumes to other volumes. You can then eject the empty, used WORM volumes and add new volumes.

Storage pools that are assigned to device classes with a device type of WORM, WORM12, or WORM14 have a default reclamation value of 100. This prevents reclamation of WORM optical media. To allow reclamation, you can set the reclamation value to something lower when defining or updating the storage pool.

Reclamation of Volumes with the Device Type of SERVER

When virtual volumes (volumes with the device type of SERVER) in a primary storage pool are reclaimed, the client data stored on those volumes is sent across the network between the source server and the target server. As a result, the reclamation process can tie up your network resources. To control when reclamation starts for these volumes, consider setting the reclamation threshold to 100% for any primary storage pool that uses virtual volumes. Lower the reclamation threshold at a time when your network is less busy, so that the server can reclaim volumes.

For virtual volumes in a copy storage pool, the server reclaims a volume as follows:

  1. The source server determines which files on the volume are still valid.
  2. The source server obtains these valid files from a primary storage pool, or if necessary, from an onsite volume (not a virtual volume) in another copy storage pool.
  3. The source server writes the files to one or more new virtual volumes in the copy storage pool and updates its database.
  4. The server issues a message indicating that the volume was reclaimed.

For information about using the SERVER device type, see Using Virtual Volumes to Store Data on Another Server.

Reclamation for Copy Storage Pools

Reclamation of primary storage pool volumes does not affect copy storage pool files.

Reclamation of volumes in copy storage pools is similar to that of primary storage pools. However, most volumes in copy storage pools may be set to an access mode of offsite, making them ineligible to be mounted. When reclamation occurs and how reclamation processing is done depends on whether the volumes are marked as offsite.

For volumes that are not offsite, reclamation usually occurs after the volume is full and then begins to empty because of file deletion. When the percentage of reclaimable space on a volume rises above the reclamation threshold, the server reclaims the volume. Active files on the volume are rewritten to other volumes in the storage pool, making the original volume available for new files.

For offsite volumes, reclamation can occur when the percentage of unused space on the volume is greater than the reclaim parameter value. The unused space includes both space that has never been used on the volume and space that has become empty because of file deletion. During reclamation, the server copies valid files on offsite volumes from the original files in the primary storage pools. In this way, the server copies valid files on offsite volumes without having to mount these volumes. For more information, see Reclamation of Offsite Volumes.

Reclamation of copy storage pool volumes should be done periodically to allow reuse of partially filled volumes that are offsite. Reclamation can be done automatically by setting the reclamation threshold for the copy storage pool to less than 100%. However, you need to consider controlling when reclamation occurs because of how offsite volumes are treated. For more information, see Controlling When Reclamation Occurs for Offsite Volumes.

Virtual Volumes:
Virtual volumes (volumes that are stored on another TSM server through the use of a device type of SERVER) cannot be set to the offsite access mode.

Reclamation of Offsite Volumes

As for volumes with other access values, volumes with the access value of offsite are eligible for reclamation if the amount of empty space on a volume exceeds the reclamation threshold for the copy storage pool. The default reclamation threshold for copy storage pools is 100%, which means that reclamation is not performed.

When an offsite volume is reclaimed, the files on the volume are rewritten to a read/write volume. Effectively, these files are moved back to the onsite location. The files may be obtained from the offsite volume after a disaster, if the volume has not been reused and the database backup that you use for recovery references the files on the offsite volume.

The server reclaims an offsite volume as follows:

  1. The server determines which files on the volume are still valid.
  2. The server obtains these valid files from a primary storage pool, or if necessary, from an onsite volume of a copy storage pool.
  3. The server writes the files to one or more volumes in the copy storage pool and updates the database. If a file is an aggregate file with unused space, the unused space is removed during this process.
  4. A message is issued indicating that the offsite volume was reclaimed.

    For a single storage pool, the server reclaims all offsite volumes that are eligible for reclamation at the same time. Reclaiming all the eligible volumes at the same time minimizes the tape mounts for primary storage pool volumes.

If you have the Tivoli Disaster Recovery Manager product, see Moving Backup Volumes Onsite.

Controlling When Reclamation Occurs for Offsite Volumes

Suppose you plan to make daily storage pool backups to a copy storage pool, then mark all new volumes in the copy storage pool as offsite and send them to the offsite storage location. This strategy works well with one consideration if you are using automatic reclamation (the reclamation threshold is less than 100%).

Each day's storage pool backups will create a number of new copy storage pool volumes, the last one being only partially filled. If the percentage of empty space on this partially filled volume is higher than the reclaim percentage, this volume becomes eligible for reclamation as soon as you mark it offsite. The reclamation process would cause a new volume to be created with the same files on it. The volume you take offsite would then be empty according to the TSM database. If you do not recognize what is happening, you could perpetuate this process by marking the new partially filled volume offsite.

One way to resolve this situation is to keep partially filled volumes onsite until they fill up. However, this would mean a small amount of your data would be without an offsite copy for another day.

If you send copy storage pool volumes offsite, it is recommended you control copy storage pool reclamation by using the default value of 100. This turns reclamation off for the copy storage pool. You can start reclamation processing at desired times by changing the reclamation threshold for the storage pool. To monitor offsite volume utilization and help you decide what reclamation threshold to use, enter the following command:

query volume * access=offsite format=detailed

Depending on your data expiration patterns, you may not need to do reclamation of offsite volumes each day. You may choose to perform offsite reclamation on a less frequent basis. For example, suppose you ship copy storage pool volumes to and from your offsite storage location once a week. You can run reclamation for the copy storage pool weekly, so that as offsite volumes become empty they are sent back for reuse.

When you do perform reclamation for offsite volumes, the following sequence is recommended:

  1. Back up your primary storage pools to copy storage pools.
  2. Turn on reclamation for copy storage pools by lowering the reclamation threshold below 100%.
  3. When reclamation processing completes, turn off reclamation for copy storage pools by raising the reclamation threshold to 100%.
  4. Mark any newly created copy storage pool volumes as offsite and then move them to the offsite location.

This sequence ensures that the files on the new copy storage pool volumes are sent offsite, and are not inadvertently kept onsite because of reclamation.

Using Storage on Another Server for Copy Storage Pools

Another resolution to this problem of partially filled volumes is to use storage on another TSM server (device type of SERVER) for storage pool backups. If the other server is at a different site, the copy storage pool volumes are already offsite, with no moving of physical volumes between the sites. See Using Virtual Volumes to Store Data on Another Server for more information.

Delaying Reuse of Reclaimed Volumes

You should delay the reuse of any reclaimed volumes in copy storage pools for as long as you keep your oldest database backup. Delaying reuse may help you to recover data under certain conditions during recovery from a disaster. For more information on delaying volume reuse, see Delaying Reuse of Sequential Access Volumes.

How Collocation Affects Reclamation

If collocation is enabled and reclamation occurs, the server tries to reclaim the files for each client node or client file space onto a minimal number of volumes. Therefore, if the volumes are manually mounted, the mount operators must:

If collocation is disabled and reclamation occurs, the server tries to move usable data to new volumes by using the following volume selection criteria, in the order shown:

  1. The volume that contains the most data
  2. Any partially full volume
  3. An empty predefined volume
  4. An empty scratch volume


[ Top of Page | Previous Page | Next Page | Table of Contents | Index ]