Browse Prior Art Database

A tool for analysis and report of multiple dump and heap files

IP.com Disclosure Number: IPCOM000218217D
Publication Date: 2012-May-28
Document File: 3 page(s) / 110K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a tool for the analysis and reporting of multiple dump and heap files. The invention provides the capability of specifying what should be examined as well as the location of the file directory. Then, the tool scans through all the thread dump files and only calculates/loads the data in which the user is interested.

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

Page 01 of 3

A tool for analysis and report of multiple dump and heap files

Server reliability over a long period time becomes a critical issue in a world where more and more functionality is dependent on server availability . However, debugging any performance or reliability degradation -over-time problem is difficult. One approach is to run a server stress test over a long time (e.g., three, days, 5 day, or longer), and monitor the server status either live or via taking snapshot .

For example, in J2EE* server reliability , scalability, stability, and availability long run tests, thread dumps and heap dumps of an enterprise application server Java Virtual Machine(JVM) under test are generated and collected periodically during the test run . These dumps are then looked at using analytic tools .

The problem is that dump files need to be loaded and examined one at a time . An administrator can load multiple files , but the end user needs to choose, select, and open these files. This process is tedious, inefficient, and only shows a snapshot of the server's health during a long run . Another problem is that loading all these files into the tools quickly uses all the memory the tool or operating system (OS) can support. Other tools that one can run during the test show the real time data and present the changing status. However, they do not serve the need of later analysis . Often, that user is interested in the server's status and wants to have a combined view on both snapshots and graphs over a period of time.

Example problem:


The analyst collected thread dump data every five minutes during a three -day stress test against a J2EE server. This test showed an over-time performance degradation of the serve...