Browse Prior Art Database

History Retention and Averaging in Performance Manager

IP.com Disclosure Number: IPCOM000119276D
Original Publication Date: 2005-Apr-01
Included in the Prior Art Database: 2005-Apr-01
Document File: 2 page(s) / 23K

Publishing Venue

IBM

Abstract

Disclosed is the algorithm that performs historical data retention and averaging for large amounts of data that is gathered at a frequent rate (every five minutes). The algorithm uses the incoming data rate as clock to maintain and retain historical averages needed by some applications. It allows such applications to scale and help maintaining large amounts of historical data needed for analysis.

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

Page 1 of 2

History Retention and Averaging in Performance Manager

The algorithm uses the incoming data rate as a clock to maintain and retain historical averages. Keeping the size of the database small is key to this invention, because smaller database results in being able to generate historical records and prune the database faster. The invention greatly reduces the amount of data stored in the database without degrading the performance analysis. The historical minimums and maximums are stored in addition to averages, so that the performance analysis results are not distorted.

The need for database purge operation is eliminated by pruning the database automatically at very frequent intervals. The history retention is done on an hourly and daily basis to handle the data collection frequency and keep the size of the database small.

Implementation: =============

The hourly, daily and weekly historical records are generated in the database for each type of statistical data. The user can specify using UI the number of latest, hourly, daily and weekly statistical records to be retained in the database. In order to generate the historical averages, a set of hourly, daily and weekly accumulator records or rows are maintained in the database for each type of statistical data collected (another alternative is to maintain these accumulator records in memory). For example, there will be an hourly, daily and a weekly statistic accumulator record generated for every entity for which historical data is retained. These accumulators contai...