Browse Prior Art Database

Java heap analysis batch process and quick browsing of analyzed data

IP.com Disclosure Number: IPCOM000168183D
Original Publication Date: 2008-Feb-29
Included in the Prior Art Database: 2008-Feb-29
Document File: 1 page(s) / 20K

Publishing Venue

IBM

Abstract

Disclosed is a system for analysis of heap dump at the time of failure and heap dump analysis format which does not require much computing resources.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 58% of the total text.

Page 1 of 1

Java heap analysis batch process and quick browsing of analyzed data

Java™ heap analysis takes longer than ever and requires more and more resources as Java heap dumps are getting larger and larger. An average analyst's machine is not powerful enough to process large Java heap dumps. Most of Java heap analysis time is spent in parsing and loading Java heap information into memory or disk. Only machines with lots of resources can be used to process large Java heap dumps. With this system, less powerful machines can download processed heap dump and browse it quickly without requiring lots of resources. The core idea is that Java virtual machine processes Java heap at the time of failure using native memory if available and stores pieces of processed Java heap into strategic locations using a special format to effectively browse analyzed information later.

1. Process Java heap dump at the time of Java heap failure by utilizing native memory if available. ( It takes 20-30 minutes only for parsing and loading 2-3 GB of Java heap dumps)
2. Print out Java heap leak suspects and store Java heap analysis data into compressed file with the following file format:

File format signature and format version (static length)

Lengths of file pointer, object size, object subtree size, start addresses of string table and entry table ( static length )

List of Java heap leak suspects ( variable length )

List of string table sorted by string (...