Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Multi-Spindle File-Placement Algorithm

IP.com Disclosure Number: IPCOM000060050D
Original Publication Date: 1986-Feb-01
Included in the Prior Art Database: 2005-Mar-08
Document File: 2 page(s) / 68K

Publishing Venue

IBM

Related People

Fierst, PE: AUTHOR [+5]

Abstract

This algorithm places disk files for optimal task and system performance on systems with multiple disk spindles. Previous attempts to place disk files involved first-fit and best-fit algorithms. The first-fit plan may result in free disk space fragmentation. The best- fit plan may result in poor input/output (I/O) balancing between spindles. The following detailed discussion will describe an improvement to the best-fit algorithm. There are two parts to this algorithm. The first part is the collection of the Disk I/O history. The second part consists of using the collected I/O history to determine file placement. Disk I/O history is defined as the sum of the number of disk operations for each disk spindle. The system maintains disk I/O activity data to be used for placement of new disk files.

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 52% of the total text.

Page 1 of 2

Multi-Spindle File-Placement Algorithm

This algorithm places disk files for optimal task and system performance on systems with multiple disk spindles. Previous attempts to place disk files involved first-fit and best-fit algorithms. The first-fit plan may result in free disk space fragmentation. The best- fit plan may result in poor input/output (I/O) balancing between spindles. The following detailed discussion will describe an improvement to the best-fit algorithm. There are two parts to this algorithm. The first part is the collection of the Disk I/O history. The second part consists of using the collected I/O history to determine file placement. Disk I/O history is defined as the sum of the number of disk operations for each disk spindle. The system maintains disk I/O activity data to be used for placement of new disk files. Three sets of data are maintained for each spindle on the machine, as pictured in Fig. 1. The work accumulator sums the disk I/O activity into a number that will be moved to the short-term data field. The short-term data places scratch files. The long-term data places permanent files. The following algorithm is shown in Fig. 2.
21. Every 6 minutes, the disk I/O activity data for each spindle is accumulated with existing data in the work accumulator fields. 22. Both short-term and long- term data fields are reset to zeros for all spindles if additional disk spindles are added. 23. Approximately every 15 days of running time, the long term data is modified to represent 25% of the actual history. This data represents system activity trends. The short-term and long-term accumulators represent different intervals of system activity to be used for placement of two different ty...