Browse Prior Art Database

Correlation of Memory Leak Condition with in Memory Session Growth

IP.com Disclosure Number: IPCOM000124891D
Original Publication Date: 2005-May-11
Included in the Prior Art Database: 2005-May-11
Document File: 1 page(s) / 22K

Publishing Venue

IBM

Abstract

Memory Leaks and the correlation of memory usage trends with in memory session growth

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 67% of the total text.

Page 1 of 1

Correlation of Memory Leak Condition with in Memory Session Growth

Current memory leak technologies are based upon the idea that you already know you have a memory leak and are trying to find it. When undetected, memory leaks cause very poor application server performance and then, finally, out of memory exceptions. Once detected ( usually after the OOM Exception) the user must then either attach an agent, or take heap dumps once they have reproduced the memory leak. Reproduction of production memory leaks is often very difficult as most shops will not be comfortable with introducing agents or taking heap dumps on their production servers.

Some of these memory leaks are caused by applications putting large objects in sessions, or an increase in the number of sessions that are stored in memory. We want to detect this scenario as a type of memory leak, alert the user, and take preventative action prior to a server crash.

Upon light weight memory detection of memory leaks, trends of total session size stored in memory can be correlated to determine if sessions are the cause of the leak. We can then take action to avoid a system crash.

In order to analyze the memory leak this solution does not involve any additional agents or heap dumps. This solution can be run in production and can automate the detection, analysis and solution to common memory leak scenarios.

When a leak is detected, keep track of the total in memory session size ( obtained when sessions are serial...