Browse Prior Art Database

Autonomic resizing of a recovery log to achieve consistent optimal performance

IP.com Disclosure Number: IPCOM000032464D
Original Publication Date: 2004-Nov-05
Included in the Prior Art Database: 2004-Nov-05
Document File: 2 page(s) / 42K

Publishing Venue

IBM

Abstract

A method by which a recovery log that employs a keypointing system can maintain optimal performance by resizing autonomically. Optimal performance is achieved by monitoring the volume of data that is being logged and the time taken to do so and resizing the log autonomically such that the maximum possible level of data throughput is maintained.

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

Page 1 of 2

Autonomic resizing of a recovery log to achieve consistent optimal performance

Disclosed is a mechanism by which a recovery log that employs a keypointing system can maintain optimal performance by resizing autonomically. Optimal performance is maintained by analysing the data which is being logged and the time taken to do so to derive the log size that will provide the highest level of data throughput.

    A common design for a recovery log is one that uses two files of identical size where, at any given time, only one file is considered to be active. Once one file becomes full the logging system switches to use the other file at which point it becomes active and the previously used file inactive. During this switching process, known as keypointing, writing to the log is suspended and any data that is live (still required) in the first file is written into the second file. Once this transfer of data has completed the second file is made active and writing to the log can resume. As the keypointing process prevents new data from being written to the log the frequency with which it occurs and the period of time that it takes has an impact on the performance of the system. The frequency at which the log keypoints is a function of the size of the file and the volume of data this is being written to it. In terms of performance the common desire is to ensure that the log's data throughput is as high and as consistent possible.

    To allow the system to automatically resize its log files in reponse to changing external parameters there are many factors that need to be considered:

The size of the data payload involved in each write operation

    The time take for each write operation
The frequency with which the log is forced
The time taken for each force operation
The frequency with which keypointing occurrs The time taken for each keypoint
The available capacity in the log immediately post-keypoint The system will be required to gather data about each of the above factors and maintain a rolling average over a...