Software Performance Engineering & Testing
I’ve added a new feature to the site: Careers
If you’re in the market, keep an eye on this page and I’ll post job openings as I receive them. If you would like to list a job opening, please send an email to job-listings@performanceengineer.com
Here are some highlights from the recent Velocity Conference:
Bill Scott wrote in his article Looks Good Works Well: Velocity Conference ‘08 Notes that the highlights for him were:
- Building Faster Pages in Firefox and Internet Explorer - This is not the same presentation as was presented & Mozilla’s is missing, hopefully they will post their slides.
- Even Faster Web Sites - This is the start of Steve’s next book on Web Site performance.
- High Performance Ajax Applications - Julien’s in depth look at tuning the world of DHTML.
- Hotmail’s Performance Tuning Best Practices - This has some great material. Great lessons learned takeaways. A favorite for me.
- IE 8 What’s Coming - Takeaway… its not Javascript engine that is slowing things down… it is rendering and layout.
- Image Optimization: How Many of These 7 Mistakes Are You Making - Crush those PNGs and more.
- Lessons Learned in Live Search Moving to and then Away from Ajax - Eric Shurman is a great speaker and full of experience. Great lessons here.
While over at Royal Pingdom they highlighted:
- Hotmail’s Performance Tuning Best Practices - Aladdin Nassar from Microsoft covers what Microsoft has learnt about website performance from Hotmail.
- High-performance Ajax Applications - Julien Lecomte from Yahoo! looks at different techniques and best practices for developing high-performance Ajax applications.
- Lessons Learned in Live Search Moving to and then Away from Ajax< - Eric Schurman from Microsoft presents the lessons learned from tuning the site performance of Live Search.
- Even Faster Web Sites - Steve Souders from Google (author of the book High Performance Web Sites and also the creator of Yslow) take a look at even more ways to speed up websites.
- Image Optimization: How Many of These 7 Mistakes Are You Making - Stoyan Stefanov from Yahoo! goes through image optimization tricks and tools that will help you speed up your website.
- Improving Netflix Performance - Bill Scott from Netflix gives plenty of insight into how Netflix improved their site performance in this interesting case study.
A couple articles about JMeter this week:
Tim posted some great JMeter tips and tricks on his blog
TSS has a new article, Master Apache JMeter and learn all its features with new book, introducing a new JMeter book:
A bad response time on a website can drive away visitors and prospective customers. To measure what a website can handle, there should be a way to simulate and analyze different load scenarios; this is where a load-testing tool like JMeter comes in. JMeter is a powerful desktop performance tool from the Apache Jakarta project, written in Java, for load-testing web pages, web applications, and other static and dynamic resources including databases, files, Servlets, Perl scripts, Java Objects, FTP Servers, and more.
For more info, check out the book’s website.
I saw this recent article about some new tools from Orbitz: InfoQ: Orbitz Open Sources Monitoring Tools ERMA and Graphite
Graphite looks particularly interesting as it has features overcoming some of the things I don’t like about RRDtool:
Graphite is a Python web application that has been developed to provide scalable storage and visualization for numeric time-series data. It receives the output from the CEP engine, which consists of data aggregated for over 70,000 metrics. A Graphite portlet was developed in order to integrate with a monitoring portal, which is based on JBoss Portal framework. The portal presents tabular and graphical views of vital system statistics and RSS feeds for alarms. Users can subscribe to feeds for particular alarm severities and/or affected applications. The Graphite web application itself is used to display visual graphs of monitoring statistics generated using RESTful URLs. Graphite data can be presented as line graphs, pie charts or raw CSV data. Graphite also provides a web-based command line interface, which power users can use to very quickly and easily create and share dashboards containing collections of related graphs.
I’ll hopefully find some time to try this out
I had a need to extract transaction response time data from a bunch of LoadRunner Analysis files, and I really didn’t want to do endless cut and paste operations from within the LoadRunner Analysis tool. I created this Python script to extract transaction response time data from the LoadRunner Analysis mdb file and output into CSV format, which can then be imported into MySQL, Excel, etc. The analysis file is a MS Access JET Database which allows the use of Python’s win32com module to access the data.
# lr_extract.py
# --------------------------------------------------------------------
# Extracts Transaction response time data from the MS Jet database file (.mdb)
# created by LoadRunner Analysis tool
#
# Requires DAO 3.6 library.
# --------------------------------------------------------------------
# Usage: python lr_extract.py lr_analysis.mdb
import sys
import string
import pythoncom
import win32com.client
const = win32com.client.constants
daoEngine = win32com.client.Dispatch('DAO.DBEngine.36')
db = daoEngine.OpenDatabase(sys.argv[1])
query = """
select
[Result].[Start Time]+[Event_meter].[End Time] as t_stamp,
[Event_map].[Event Name] as txn_name,
[Event_meter].[Value] as resp_time,
[Result].[Result Name] as result_name
from
Result,
Event_map,
Event_meter
where
[Event_map].[Event Type] = 'Transaction' and
[Event_meter].[Event ID] = [Event_map].[Event ID]
order by [Result].[Start Time]+[Event_meter].[End Time]
"""
rs = db.OpenRecordset(query)
print("timestamp,transaction,response time,result name")
while not rs.EOF:
print("%s,%s,%s,%s" % (rs.Fields[0],rs.Fields[1],rs.Fields[2],rs.Fields[3]) )
rs.MoveNext()
"Pylot is an open-source performance and scalability testing tool. It uses HTTP load tests so that you can plan, benchmark, analyze and tweak performance. Pylot requires that you have Python installed on the server - but you don’t need to know the language, you use XML to create your testing scenarios."
The Grinder is:
The Grinder is a JavaTM load testing framework that makes it easy to run a distributed test using many load injector machines.
Until recently, I had been reluctant to investigate The Grinder because I was unfamiliar with Jython/Python, and hesitant to dig into a new language. I was wrong.
Not only am I happy to have discovered that Python and Jython are not too difficult to pick up, but The Grinder is quite a flexible tool that has a lot of potential and I would say a nearly capable replacement for LoadRunner from a load generation perspective.
In this article I’ll review how to setup The Grinder and use Eclipse to edit your scripts.
Quick tip: With prevalence of larger monitors these days, it is nice to have charts and graphs that take advantage of the extra space. If you want to customize the size of the charts in the LoadRunner HTML reports, change these settings in C:\Program Files\Mercury\LoadRunner\bin\dat\GeneralSettings.txt:
[HTML report] ChartWidth=1024 ChartHeight=576
If you’ve been to this site before, you may have noticed the new look. For various reasons, I’ve converted the site from drupal to Wordpress. Everything seems to be working, and hopefully most of the old links will redirect to their new locations.
One thing that is no longer here are the forums. They weren’t too active, but if you have a question feel free to add a comment here or find me on Twitter at http://twitter.com/perfeng
Thanks for visiting, come back often!