Browse Prior Art Database

Method to Improve Process Startup Performance by Using Pre-processing Technique for Reading Static XML Files

IP.com Disclosure Number: IPCOM000097710D
Original Publication Date: 2005-Mar-07
Included in the Prior Art Database: 2005-Mar-07
Document File: 2 page(s) / 106K

Publishing Venue

IBM

Abstract

Disclosed program provides a technique to efficiently read static XML files and in turn improve the process startup performance

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 52% of the total text.

Page 1 of 2

Method to Improve Process Startup Performance by Using Pre -processing Technique for Reading Static XML Files

Problem

Typically, the configuration files are stored in XML format and in a large product like IBM WebSphere, there are a lot of XML configuration files that are static (immutable during the runtime). The process of parsing an XML is expensive and memory intensive. As a result it affects the process startup performance since most of the configuration files are read during the startup time.

Solution

To reduce the impact of XML parsing during the process startup the XML files are pre-processed during the compile time and converted into an intermediate format (Java class or Java serialized file). The runtime now reads the intermediate format which is faster than parsing the XML file. The results indicate a significant reduction in CPU usage when using the preprocessing technique (from 2% to 0.5% of CPU cycles -- about 75% improvement). Figure 1 describes this technique in detail.

Problem and solution scenario - 1

IBM WebSphere Performance Monitoring Infrastructure (PMI) XML template defines the statistics specification for the component. This template is used by the PMI component and it doesn't change during the runtime. In 5.x, this file is parsed and validated every time the server

1

[This page contains 1 picture or other non-text object]

Page 2 of 2

is started. As the number of PMI templates increased in 6.0, this approach of parsing during server startup is proving to impact the server startup time.

In 6.0 the templates will be parsed and validated once during the build time and a Java look-up class will be generated. Each component will be able to generate a look-up class for all the templates in the component using an ANT task. The component should then register the generated look-up class with the PMI during the runtime. The PMI runtime will use the generated look...