Planning the storage pool hierarchy

Edit online

Plan the storage pool hierarchy to ensure that data is migrated daily from disk to tape. The migration releases space on the disk device and moves the data to tape for long-term retention. In this way, you can take advantage of the scalability, cost efficiency, and security features of tape storage.

Before you begin

The storage pool hierarchy helps to manage the flow of data. To understand the data flow, review Figure 1.
Figure 1. Tape solution
The image shows the server using disk-to-disk-to-tape backup. Storage pool staging is used, in which client data is initially stored in disk storage pools and then migrated to tape storage pools.
The following steps correspond to the numbers in the figure:
  1. The server receives data from clients (applications, virtual machines, or systems) and stores the data on primary storage pools. Depending on the client type, the data is stored on a primary storage pool on disk or tape.
  2. The data on disk and tape is backed up to a copy storage pool on tape.
  3. Data in the primary storage pool on disk is migrated daily to the primary storage pool on tape.
  4. Data from the copy storage pool on tape is moved offsite to support long-term retention and disaster recovery.

Procedure

To plan the storage pool hierarchy, answer the following questions:
  1. Which clients should back up data to disk, and which clients should back up data to tape?
    • The preferred method is to back up clients that host large objects, such as databases, to tape.
    • The preferred method is to back up all other clients to disk.
    • Virtual machine (VM) clients can be backed up to disk or tape. The preferred method is to back up a VM client to a separate disk storage pool, which is not migrated to tape. If you must migrate a VM client to tape, create a smaller disk storage pool to hold the VMware control files. This smaller disk storage pool cannot be allowed to migrate to tape. For more information about backing up a VM client to tape, see Tape media guidelines and and IBM Tivoli Storage Manager guest support for Virtual Machines and Virtualization.
    Tip: If many clients must back up data to a single storage pool, consider using a storage pool on disk because you can specify many mount points. You can specify a maximum value of 999 for the MAXNUMMP parameter on the REGISTER NODE command.
  2. What are the considerations for specifying the capacity of disk-based storage pools?

    At minimum, plan enough capacity to store data from a single day of backup operations. The preferred method is to plan enough capacity to store data from two days' worth of backup operations and add a 20% buffer.

  3. What are the considerations for specifying the device class for the disk-based storage pool?

    The preferred method is to specify a FILE device class. Set the MOUNTLIMIT parameter to 4000. Also, ensure that the node has a sufficiently high number of mount points, which you can specify by using the MAXNUMMP parameter on the REGISTER NODE command.

  4. Should data deduplication be specified for the disk storage pool?

    No, because the data is stored on disk for only one day before the data is migrated to tape.

  5. Should automatic migration of data be specified based on a migration threshold?

    No. Instead, plan to schedule daily migration by using the MIGRATE STGPOOL command. (To prevent automatic migration based on the migration threshold, specify a value of 100 for the HIGHMIG parameter and 0 for the LOWMIG parameter when you issue the DEFINE STGPOOL command.)

  6. Should a migration delay be specified?

    The preferred method is to specify migration from disk to tape daily, and not specify a migration delay, which requires additional planning. For more information about migration delays, see Migrating files in a storage pool hierarchy.

  7. How can the number of tape drives be calculated?
    1. Determine the native data transfer rate of the drive by reviewing the manufacturer's documentation. To obtain an estimate of the sustained data transfer rate in your storage environment, subtract 30% from the native data transfer rate.
    2. Calculate the required rate of data ingestion by the server. Then, divide that figure by the sustained data transfer rate of a single tape device. The result is the minimum number of drives to support data ingestion.
    3. Calculate the number of mount points that are required by clients that back up data to tape, including those clients that use multiple sessions. You can distribute the mount points over the backup window, keeping in mind that clients are likely backing up large objects, which might use most of the window.
    4. Calculate the performance requirements and mount points that are required for maintenance tasks, such as disk-to-tape migration and tape-to-tape copies. By backing up data to tape, you can avoid migration processing, but making tape-to-tape copies will double the tape drive requirement.
    5. Calculate the number of additional drives that might be required, for example:
      • If a tape drive malfunctions, the issue impacts the number of available mount points and the ingestion rate. Consider provisioning spare drives. For example, if you require five tape drives for normal operations, consider provisioning two spare drives.
      • Restore and retrieve operations might require additional tape drives if you plan to run the operations simultaneously with data ingestion and maintenance operations. If necessary, provision additional tape drives and ensure that they are unused when you start the restore or retrieve operations.
  8. What alternatives are available for optimizing restore operations?
    You can use collocation to improve system performance and optimize data organization. Collocation can reduce the number of volumes that must be accessed when a large amount of data must be restored:
    • For disk-based storage pools, the preferred method is to use collocation by node. The server stores the data for the node on as few volumes as possible.
    • For tape-based storage pools, the preferred method is to use collocation by group. Collocation by group results in a reduction of unused tape capacity, which allows for more collocated data on individual tapes.
    For more information about collocation, see Optimizing operations by enabling collocation of client files.
    If you are an experienced system administrator, you might plan additional actions to optimize restore operations. See Optimizing restore operations for clients, File backup techniques, and MOVE NODEDATA (Move data by node in a sequential access storage pool).