lundi 13 avril 2009

Are the data from the GoogleApp Engine Dashbord valid?

The DashBord of GoogleApp Engine presents interesting results, but what is the truth behind these data?


I started by deploying a simple Java EE (JSP / Struts / Strus menu) and déploing it in the GoogleApp Engine cloud at http://kbdsoft.appspot.com/


This section aims to: Check the values presented by the DashBord of GoogleApp Engine.

To Check the validity of the values presented by the DashBord of GoogleApp Engine, we use a simulation tool in the classical world of performance studies: the famous Jmeter

Install a simulation tool load

Download Jmeter from http://jakarta.apache.org/jmeter/
Launch Jmeter Script and create a simulation application execution

Beware, the testing was performed on a laptop connected to the Internet with a limited speed, ADSL 1Mb (theoretical), it has no intrinsic value to measure the performance of the platform, but will compare the results .)

Run a simulation of 20 concurrent virtual users (20 Threads)

The scenario is executed 100 times.

http://kbdsoft.appspot.com/

http://kbdsoft.appspot.com/pagesManagement.do

http://kbdsoft.appspot.com/menuManagement.do

http://kbdsoft.appspot.com/user.do

http://kbdsoft.appspot.com/audit.do

http://kbdsoft.appspot.com/rolesManagement.do

http://kbdsoft.appspot.com/index.do

We note that at the beginning of the simulation


We note that at the beginning of the simulation

At the end of simulation the graph Request / Second becomes

The GoogleApp Engine’s Dashbord allows you to track the behavior of the application: it is the implementation of the concept SAM (Service Activity Monitoring, for monitoring SLAs stipulated in the contract for the paid version)

Thus, after the full execution of the script, the results presented by the GAE's Dashbord are summarized as follows:

On the other hand, after the end of the execution of the script, the results presented by Jmeter are summarized as follows :


The results of Jmeter Vs results GoogleApp Engine

There is a good concordance of results (although it sure is a short and simulation for more detailed results, it must be remembered that performance measures should follow a statistical study and use the laws of large numbers)



Jmeter

client vue

GAE DashBord

Google's serveurs vue

http://kbdsoft.appspot.com/

1.135

1.2

http://kbdsoft.appspot.com/pagesManagement.do

2.265

2.4

http://kbdsoft.appspot.com/menuManagement.do

2.275

2.4

http://kbdsoft.appspot.com/user.do

3.411

3.6

http://kbdsoft.appspot.com/audit.do

1.142

1.2

http://kbdsoft.appspot.com/rolesManagement.do

1.144

1.2

http://kbdsoft.appspot.com/index.do

1.147

1.2

Total

12.59

13.5

The results shown by Google are validated by the results of Jmeter

We can rely on the DashBord.
Conclusion:
For this first test, I "fell in love": it took me half an hour to deploy an old application (simple) Struts based on the Google cloud

No changes were necessary!
(N: there was no transaction, database, Spring, Hibernate ... )
Dashboard of
GoogleApp Engine allows to monitor the activity of the application and see the logs and the response time.


Working with GoogleApp Engine is :

  • Easy
  • Lets recycle "reflexes" of Java developers
  • Lets recycle "old" Java
  • The Dashboard provides information close to the reality and verifiable

Le cloud coumputing en marche ...



0 commentaires :

Enregistrer un commentaire

Architecte SOA & Professionnel Open Source Headline Animator

 
Khaled BEN DRISS
Cloud Computing, SOA et Web 2.0 : Des sujets techniques sur SOA et l'Open Source : de Java & .Net, PHP5, Symfony, à SaaS / PaaS en passant par Azure, google appengine, le BPM, la Modélisation et d'autres sujets du coté du serveur et cloud computing.