AppDynamics ? What, Why & How

If you ask me what is "AppDymamics", in my own words, it is real-time web application profiler and monitoring tool, reason i called it in such way is the features i used most in the free version. 

Actually, AppDynamics call itself Application Performance Management (APM) product and has more features then i blogged here. This post is just a jump-start and personal experience of AppDymamics Lite (vs Commercial version) in my development experience.

Two most important reasons made me to try it out: 

- Easy to install

My team is so busy in development and no time to setup individual application profiler on their IDEs. 

With few minutes' effort, i setup one from our DEV server and allow all my developers to use with no wait. 

- Easy to use 

Very intuitive GUI and allow to drill down all the way to JDBC and SQL query with detail call stack (call graph) and time spending.

No training required, just follow the color code and double-click and click, click until I found the root cause in minutes.

So, When should use application Profiler instead of AppDynamics? Here is a short answer:    

WHEN TO USE A PROFILER	WHEN TO USE APPDYNAMICS LITE
You need to troubleshoot high CPU usage and high memory usage	You need to troubleshoot slow response times, slow SQL, high error rates, and stalls
Your environment is Development or QA	Your environment is Production or performance load-test
15-20% overhead is okay	You can't afford more than 2% overhead

Now, let me walk you through how to start & use the AppDynamics


- Download the appdynamics and install it in 4 steps. 

- Open web browser and browse through the slow web pages or the processes you intend to monitor and profile

- Open web browser and login AppDynamics (Be patient, if you don't see any data. AppDynamics need a short while to build and load up the analytic info) 

^Landing page will show list of recent transactions (web request) and highlight the response time in different color code depending on the settings (slow, very slow and stalling) which you can change at your own will. 

- Double-click the page name (URL) in red color and until you see the full call graph & timing of each spending.


    * First, it will drill down to all sampling of the same transaction (web request) 

   * Chose one of them and double-click to drill down to call graph

    ^you can filter the graph by method timing spending and class type (servlet, pojo, spring bean, ejb and etc)  

    

   *Click and click ... until you see something you are familiar 

    ^Take note of color code, don't confuse the total time spending (ORANGE) with method self time spending (BLUE)

As recent transactions are not good enough, we need to check out long transaction in past 30 minutes or 1 hours. Never mind, go and visit SnapShot page.

   * Click on SnapShot Tab

   ^ Too much info in the page, never mind, enable filter to view less and more concerned transactions.

    * Double-click the long transaction to pull out the call graph

   ^ There is JDBC call in this transaction

    * Right click on slow method to further expend the graph

   * Click on JDBC to check out list of executed SQL & time spending 

    * Scroll Up & Down and check what are the SQL and how much they spend in the total timing. 

    

   * Save the call graph for offline or offsite study by other developers  

    * Save to local disc and sent it to developers for further study. 

    ^ This file also includes list of executed SQL and timing of each.   

That is all, it is sweet & easy experience. In fact, i have put it up behind an Apach Server so that our offsite team can access and check out root cause of long transaction.