Browse Prior Art Database

Minimizing Storage Device Contention in a Multi-host, Multi-tasking Environment

IP.com Disclosure Number: IPCOM000108514D
Original Publication Date: 1992-Jun-01
Included in the Prior Art Database: 2005-Mar-22
Document File: 4 page(s) / 127K

Publishing Venue

IBM

Related People

Hsiao, Y: AUTHOR [+3]

Abstract

Data Facility Hierarchical Storage Manager (DFHSM) is the IBM inactive data manager for the MVS operating system on mainframe computer systems. In its space management function, DFHSM moves low-activity data sets from user owned, DFHSM-managed DASD volumes to either DASD or tape volumes owned by DFHSM.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Minimizing Storage Device Contention in a Multi-host, Multi-tasking Environment

       Data Facility Hierarchical Storage Manager (DFHSM) is the
IBM inactive data manager for the MVS operating system on mainframe
computer systems.  In its space management function, DFHSM moves
low-activity data sets from user owned, DFHSM-managed DASD volumes to
either DASD or tape volumes owned by DFHSM.

      In releases prior to R 2.6.0, DFHSM used a DASD target volume
selection algorithm based upon choosing the DASD volume with the most
available free space.  With the addition of multi-tasking migration
in DFHSM R 2.6.0, a new algorithm was introduced to take best
advantage of the new environment.

      In a multi-tasking environment where DASD volumes are shared
among various tasks, system performance is affected by the number of
simultaneous users on each volume.  If free space is the primary
criteria used for device selection, tasks tend to "gang up" on DASD
volume that has the most free space at any particular moment in time.
To eliminate this effect, a record can be kept of the free space
available on, and the number of tasks using, each device. Each task
needing to store data will check this record and select a device in
the set of devices with the fewest number of current users.  The
device selected will have the most free space in this least-users
set.
GENERAL DESCRIPTION
1.  At product start-up a common record is created and updated with
the free space on each storage device.
2.  The common record is copied to a host-specific in-memory record
for each host in the system.  Besides the free space, this in-memory
record also contains fields for the number of users of each storage
device.
3.  For each device selection the in-memory record is scanned for the
entry having the least number of users and the most free space.
4.  When a device is selected, the number of users is incremented in
the in-memory record.
5.  When the data set is allocated, the free space in the in-memory
record is decremented by the size of the data set allocation.
6.  When a task is finished allocating and writing a data set, the
number of users is decremented.
7.  When a data set is scratched, the in-memory free space for the
affected device is incremented.
8.  Periodically the in-memory and common records are serialized and
combined to bring all records up-to-date for device free space
information.
EXPANDED DESCRIPTION
1.  Create In-Memory and Common Records at Startup
* The user supplies a list of all product-owned storage devices to
include in the common record.
* The product calculates for each supplied storage device the number
of free space elements by reading the volume's table of contents.
* The product creates or updates the common record so it contains an
entry for each device.  The entry contains the newly generated amount
of free space and the storage device label (VOLID).
* The common record is...