Posted: January 12th, 2010 | Author: TnT Admin | Filed under: Analyze | Tags: Analyze, Bottleneck, database, MS SQL, SQL Statements, Stored Procedures | No Comments »
After the tune-able parameters are changed for optimal performance, your system still fails miserably with a poor response time. The most likely step you should take is to do a deep diagnostics on the system. Break the system up into different components such as application server where diagnostics using probes is required, database servers where SQL statements and stored procedures become the next to be scrutinized, etc… Read the rest of this entry »
Posted: January 4th, 2010 | Author: TnT Admin | Filed under: Tools | Tags: Add-Ons, Analyze, Bottleneck, firefox | No Comments »
YSlow, another popular client-side performance profiling tool for web applications, is an add-on for Firefox. With YSlow, you will be able to determine how fast (or slow) your web page is loading, breaking down to the individual components on the web page. One thing to note, YSlow requires another add-on, Firebug to be installed in Firefox before you can install it. With YSlow you can use to determine the following website performance problems: Read the rest of this entry »
Posted: October 24th, 2008 | Author: TnT Admin | Filed under: How-Tos | Tags: Analyze, Graphs | 6 Comments »
In LoadRunner, the percentile graph is a great graph for understanding how well the system is performing in terms of transaction response time from an overall perspective in a load test. How is the percentile graph derived? The mathematical explanation/formula can be found in Wiki, Percentile rank. Read the rest of this entry »
Posted: July 31st, 2008 | Author: TnT Admin | Filed under: Analyze | Tags: Analyze, WAS | No Comments »
Taken from the IBM Public Library for Websphere 4.x (a little out-dated). In this featured article for WebSphere 4.x, the concept is applicable to most J2EE application in determining if java.lang.OutOfMemoryError is due to allocation of large heap objects which is the cause of heap fragmentation. Heap fragmentation can be detected when there is a (a) high amount of free heap and a (b) high amount of memory in heap while a (c) “totally out of heap space message” occur. Techniques on how to resolve the problem such as coding it to prevent heap fragmentation, diminishing the effect of potential heap fragmentation are discussed in the article, etc. To get the article, click on the following link: Read the rest of this entry »
Posted: April 21st, 2008 | Author: TnT Admin | Filed under: Analyze | Tags: Analyze, Disk | No Comments »
When you are running a load test and you may have encounter that the % Disk Time exceeds 100% in Controller or when you are analyzing the graphs in Analysis. Over 100%? Weird right? And when you open perfmon to verify, and it’s still more than 100%. Is it a bug of Windows? What’s the cause behind this? Read the rest of this entry »