Tivoli Header

Administrator's Guide


Partitioning IBM 349X Libraries in a LAN

Tivoli Storage Manager uses the capability of the 349X library manager, which allows you to partition libraries that reside on a LAN between multiple Tivoli Storage Manager server applications. Library partitioning differs from library sharing in that there are no Tivoli Storage Manager library managers or clients. Each server has the same access to the library. Each drive that resides in the library is assigned to one server. Each server can then only access those drives it has been assigned. Tape volume categories must also be assigned to each server that has access to the library. As a result, library partitioning does not allow dynamic sharing of drives or tape volumes since they are pre-assigned to different servers using different names and category codes.

In the following example, an IBM 3494 library containing four drives is attached to a Tivoli Storage Manager server system named ASTRO and to another Tivoli Storage Manager server system named JUDY.

Set up the Device on the Server System

You must first set up the 349X library on the server system. This involves the following tasks:

  1. Set the 349X Library Manager Control Point, or LMCP. This procedure is described in the IBM SCSI Tape Drive, Medium Changer, and Library Device Drivers Installation and User's Guide for AIX.
  2. Physically attach the devices to the server hardware.
  3. Install and configure the appropriate device drivers for the devices.
  4. Determine the device names that are needed to define the devices to Tivoli Storage Manager.
For details, see Attaching an Automated Library Device and Installing and Configuring Device Drivers.

Define the Devices to Tivoli Storage Manager ASTRO

  1. Define the 3494 library named 3494LIB:
    define library 3494lib libtype=3494 privatecategory=400 scratchcategory=600
    
    The PRIVATECATEROY and SCRATCHCATEGORY are set differently from the default settings. See Categories in an IBM 349X Automated Library.
  2. Define the path from the server, ASTRO, to the library:
    define path astro 3494lib srctype=server desttype=library 
     device=/dev/lmcp0
    

    The DEVICE parameter specifies the device special file for the LMCP.

    See Defining Libraries and SCSI Libraries. For more information about paths, see Defining Paths.

  3. Define the drives that are partitioned to server ASTRO:
    define drive 3494lib drive1
    define drive 3494lib drive2
    
  4. Define the path from the server, ASTRO, to each of the drives:
    define path astro drive1 srctype=server desttype=drive library=3494lib 
     device=/dev/rmt0
    define path astro drive2 srctype=server desttype=drive library=3494lib 
     device=/dev/rmt1
    

    The DEVICE parameter specifies the device special file for the drive. In this example, each drive is given a Tivoli Storage Manager name that is unique to the device special file. For more about device names, see Determining Device Special File Names. For more information about paths, see Defining Paths.

  5. Classify drives according to type by defining Tivoli Storage Manager device classes. We recommend that you use FORMAT=DRIVE as the recording format only if all the drives associated with the device class are identical. For example, to classify the two drives in the 3494LIB library, use the following command to define a device class named 3494_CLASS:
    define devclass 3494_class library=3494lib devtype=3590 format=drive
    

    See Defining and Updating Device Classes for Tape Devices.

  6. Verify your definitions by issuing the following commands:
    query library
    query drive
    query path
    query devclass
    

    See Requesting Information About Libraries, Requesting Information about Drives, Requesting Information about a Device Class, and Requesting Information About Paths.

  7. Define a storage pool named 3494_POOL associated with the device class named 3494_CLASS.
    define stgpool 3494_pool 3494_class maxscratch=20
    

    Key choices:

    1. Scratch volumes are empty volumes that are labeled and available for use. If you allow scratch volumes for the storage pool by specifying a value for the maximum number of scratch volumes, the server can choose from the scratch volumes available in the library, without further action on your part. If you do not allow scratch volumes, you must perform the extra step of explicitly defining each volume to be used in the storage pool.

    2. Collocation is turned off by default. Collocation is a process by which the server attempts to keep all files belonging to a client node or client file space on a minimal number of volumes. Once clients begin storing data in a storage pool with collocation off, you cannot easily change the data in the storage pool so that it is collocated. To understand the advantages and disadvantages of collocation, see Keeping a Client's Files Together: Collocation and How Collocation Affects Reclamation.

    For more information, see Defining or Updating Primary Storage Pools.

Define the Devices to Tivoli Storage Manager JUDY

  1. Define the 3494 library named 3494LIB:
    define library 3494lib libtype=3494 privatecategory=112 scratchcategory=300
    
    The privatecategory and scratchcategory are defined differently that the first server's definition. See Categories in an IBM 349X Automated Library.
  2. Define the path from the server, JUDY, to the library:
    define path judy 3494lib srctype=server desttype=library 
    device=/dev/lmcp0
    

    The DEVICE parameter specifies the device special files for the LMCP.

    See Defining Libraries and SCSI Libraries. For more information about paths, see Defining Paths.

  3. Define the drives that are partitioned to server JUDY:
    define drive 3494lib drive3
    define drive 3494lib drive4
    
  4. Define the path from the server, JUDY, to each of the drives:
    define path judy drive3 srctype=server desttype=drive library=3494lib
     device=/dev/rmt2
    define path judy drive4 srctype=server desttype=drive library=3494lib
     device=/dev/rmt3
    

    The DEVICE parameter gives the device special file for the drive. In this example, each drive is given a Tivoli Storage Manager name that is unique to the device special file. For more about device names, see Determining Device Special File Names. For more information about paths, see Defining Paths.

  5. Classify drives according to type by defining Tivoli Storage Manager device classes. We recommend that you use FORMAT=DRIVE as the recording format only if all the drives associated with the device class are identical. For example, to classify the two drives in the 3494LIB library, use the following command to define a device class named 3494_CLASS:
    define devclass 3494_class library=3494lib devtype=3590 format=drive
    

    See Defining and Updating Device Classes for Tape Devices.

  6. Verify your definitions by issuing the following commands:
    query library
    query drive
    query path
    query devclass
    

    See Requesting Information About Libraries, Requesting Information about Drives, and Requesting Information about a Device Class.

  7. Define a storage pool named 3494_POOL associated with the device class named 3494_CLASS.
    define stgpool 3494_pool 3494_class maxscratch=20
    

    Key choices:

    1. Scratch volumes are empty volumes that are labeled and available for use. If you allow scratch volumes for the storage pool by specifying a value for the maximum number of scratch volumes, the server can choose from the scratch volumes available in the library, without further action on your part. If you do not allow scratch volumes, you must perform the extra step of explicitly defining each volume to be used in the storage pool.

    2. Collocation is turned off by default. Collocation is a process by which the server attempts to keep all files belonging to a client node or client file space on a minimal number of volumes. Once clients begin storing data in a storage pool with collocation off, you cannot easily change the data in the storage pool so that it is collocated. To understand the advantages and disadvantages of collocation, see Keeping a Client's Files Together: Collocation and How Collocation Affects Reclamation.

    For more information, see Defining or Updating Primary Storage Pools.


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