Browse Prior Art Database

A table driven methodology for vetting and ordering tape dataset placement and access.

IP.com Disclosure Number: IPCOM000021753D
Original Publication Date: 2004-Feb-05
Included in the Prior Art Database: 2004-Feb-05
Document File: 2 page(s) / 34K

Publishing Venue

IBM

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

Page 1 of 2

A table driven methodology for vetting and ordering tape dataset placement and access.

This article describes :

a resource manager that manages access, for either input or output, to datasets on mountable media. Further, data on the media is assumed to be serially accessible so that a degree of optimization is undertaken to provide greater efficiency in accessing datasets on the media. a methodology for dynamically allocating the mountable media based upon information returned by the resource manager, once the queues are built, a methodology for managing these dynamically allocated, mountable media by the caller of the resource manager.

Input from JCL and/or Templates describing datasets associated with objects is vetted and the objects are placed in queues such that when objects from any given queue are processed serially, mounting and demounting of media, as well as positioning on the media is minimized. Further, it is guaranteed that no object from one queue will ever need the mountable media referenced by an object in another queue, such that the queues can be processed independently of each other subject to constraints such as the number of available addresses for mounting the media, or virtual storage for spawning subtasks.

If an indication of the maximum number of dynamically allocatable drives is not provided the resource manager will determine the number.

Drives on which the mountable media are mounted can be allocated via JCL or Dynamic Allocation. Only those drives that have been dynamically allocated are managed. Drives that are managed are kept allocated as long as they are needed and can service requests. Specifically, requests that service stacked datasets are optimized, where stacked datasets are multiple datasets on the same volume. Further, when a new drive is required to be allocated any already allocated drives that can be reused, will be reused before a new allocation is made, this minimizing the number of drives that are required to service the total request.

Building the queues

There are 3 types of queues:

Unordered:

This queue contains all objects that can be processed completely independently of any other object in any queue. Such objects would be those on media that can be multiply accessed, or serially accessible...