Browse Prior Art Database

Write-Modified Index Blocks During System Idle Time

IP.com Disclosure Number: IPCOM000042526D
Original Publication Date: 1984-May-01
Included in the Prior Art Database: 2005-Feb-04
Document File: 3 page(s) / 38K

Publishing Venue

IBM

Related People

Aiken, JA: AUTHOR

Abstract

The disclosed Storage Access Method (SAM) for a word processing system uses a priority request scheme and Auto-Requests to cause modified memory-resident index blocks to be written to a storage volume during times when the SAM is not servicing system application requests. The time to complete an application service request is not significantly delayed, since servicing an Auto-Request is not initiated until all queued application requests have been completed. Initiation of the next application service request while processing an Auto-Request is delayed at most until the current index block has been processed. For normal system operations, Checkpoint and document revision termination (close) requests complete much quicker, since there are no modified index nodes which need to be written to the storage volume.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 43% of the total text.

Page 1 of 3

Write-Modified Index Blocks During System Idle Time

The disclosed Storage Access Method (SAM) for a word processing system uses a priority request scheme and Auto-Requests to cause modified memory- resident index blocks to be written to a storage volume during times when the SAM is not servicing system application requests. The time to complete an application service request is not significantly delayed, since servicing an Auto- Request is not initiated until all queued application requests have been completed. Initiation of the next application service request while processing an Auto-Request is delayed at most until the current index block has been processed. For normal system operations, Checkpoint and document revision termination (close) requests complete much quicker, since there are no modified index nodes which need to be written to the storage volume. Documents for the present word processing system are stored on magnetic media storage volumes, either diskettes or an optional internal hard disk file. As in most stand-alone word processing systems, the time it takes to complete an access or update to a portion of a document in a storage volume is one of the factors in determining operator productivity. The most significant factor in this respect is the number of I/O operations issued to the diskette or hard disk file. The data in a storage volume is organized into data sets. The data set index is tree-structured, with the first level being called the Root Node. Data sets are accessed through the Volume Table of Contents (VTOC) or the volume index, which is also tree- structured. The first level of index is called the Anchor. As data sets grow or become smaller, volume sectors may be allocated to the data set or may be returned to the pool of available sectors. The Media Allocation Map indicates for each sector whether that sector is available for future allocation or already allocated. The locations in the storage volume of the Anchor, Media Allocation Map, and data set Root Nodes are not pre-defined or pre-allocated. The Anchor and Media Allocation Map are allocated to the first available sectors in the volume when the volume is initialized. Data set roots are allocated to the first suitable extent at the time the data set is created. In order to improve performance of access and updating, the Anchor and Media Allocation Map are kept in a system memory at all times when the particular volume is available (mounted). Referring to the drawing, the Storage Unit Control Block (UCS) contains space for the Anchor and Media Allocation Map. There is one UCS for each diskette drive or hard disk file attached to the system. For each data set which is to be accessed, the system reserves a Storage Access Control Block (SACB). Since all accesses or updates must begin with an index search, and since all such searches start with the Root Node, a copy of the Root Node is stored in the SACB. For small data sets, the Root Node is the entire...