Performance Sizing and Forecast for Oracle Databases
From OraWiki
Linear regression analysis for Oracle, SAP and other performance experts
Contents |
Author
| (c) 1992-2007 by Gerald Roehrbein | ||
|---|---|---|
| Author | ||
| Name | Gerald Roehrbein | |
| http | www.oraforecast.com | |
| Gerald.Roehrbein@OraForecast.com | ||
| Release | 0.99 | |
| Created | 2008.01.24 | |
| Last update | 2009.01.24 | |
| Changes | New Content | |
| Estimated duration of installation | 5 minutes | |
| License | GPL 2.0 | |
Purpose
A small JAVA regression analysis tool for Oracle databases. This tool helps with
- Sizing
- Forecast
- Measurement of performance optimization efficiency
Installation
Usage
List of functions
If you found this page and waited for Java applet download than you will see two Java dialogs.
- Config Dialog
- Simple Data Plotter
To access sample data you have to change the content of host field in Config Dialog to www.oraforecast.com and the content of SID field to XE. After this you have to press the apply button at the bottom of the the Config Dialog.
After this step the Plotted SID list box is filled with sample data of system DEMO and SAMPLE. The Simple Data Plotter window will be filled automatically with sample data of the DEMO system.
Now you can start analysis of sample data.
This works at this page since 2008.01.26.
Recommendations
Example(s)
This is a place holder for the Java applet SimplePlotter.
The next image shows a screenshot of SimplePlotter. You could start a live demo using the link below the image. If you can not run the live demo or it is not stable than it's most likely that you have a firewall or other network related problems.
If you want to run a live demo of Simple Plotter than click at Simple Plotter Live Demo
Security related issues
How to make this feature as secure as possible
Source code of this extension
I will publish source code and java compiled code of the analysis tool "SimplePlotter" and the "Gatherer" process at this location.
So you should be able to use this tool at any location. The gatherer process have to be installed at the server where you want to gather data. The process gathers CPU, IO using operating system fucntions and database usage like "consistent gets" in a 10 minute time intervall.
If you have this data you can calculate which resource will limit systems throughput and when the system will reach it's limit and if you've done some tuning of your instance, SQL or datamodel than you will see, that the point in time when your system will become stale will change. If your work was efficient than this point will move to the far future.
Since I've developed this at night in a hotel the code is not very elegant but it's simple to understand. So please don't blame me as a bad Java programmer. I've developed this as I had a problem to get a fast and cheap solution. Most forecast utilities are very expensive and most managers do not understand why it makes sense to have something like that.
Download of Regression Analysis module
Documentation
Common Performance Ideas/ Considerations
