onTune Case Study sep 2012

2
Easiest Way to Manage Your Critical Systems Customer Case Study Copyright2012TeemStone Pty. Ltd. All Right Reserved OnTune Case Study September. 2012 Case For an internet shopping mall, a system operator developed an interactive table with flash, installed and operated the system. But the customer faced frequent system errors, and finally was not able to provide the service. Issue After the internet shopping mall opened, the new developed flash program was executed for 2 hours, and then it became to be crashed steadily so that it couldn’t provide continual services. - There was no way to figure out the system status for the Operating System (Windows). : The System Operator considered implementing data logging through ‘perfmon’ provided by windows, but he couldn’t decide which value should be done logging above lots of performance counters. Even though he completed logging, he faced difficulty for analyzing huge data which didn’t stand out well - In case of making down the interactive table for Application Debugging, it is impossible to do requested service process, neither able to see the process of actual applications, which means the System operator cannot figure out the root cause of the problem.

description

 

Transcript of onTune Case Study sep 2012

Page 1: onTune Case Study sep 2012

Easiest Way to Manage Your Critical Systems

Customer Case Study Copyright2012ⓒTeemStone Pty. Ltd. All Right Reserved

OnTune Case Study September. 2012

� Case

For an internet shopping mall, a system operator developed an interactive table with flash, installed and operated the system. But the customer faced frequent system errors, and finally was not able to provide the service.

� Issue

After the internet shopping mall opened, the new developed flash program was executed for 2 hours, and then it became to be crashed steadily so that it couldn’t provide continual services.

- There was no way to figure out the system status for the Operating System (Windows).

: The System Operator considered implementing data logging through ‘perfmon’ provided by windows, but he couldn’t decide which value should be done logging above lots of performance counters. Even though he completed logging, he faced difficulty for analyzing huge data which didn’t stand out well

- In case of making down the interactive table for Application Debugging, it is impossible to do requested service process, neither able to see the process of actual applications, which means the System operator cannot figure out the root cause of the problem.

Page 2: onTune Case Study sep 2012

Easiest Way to Manage Your Critical Systems

Customer Case Study Copyright2012ⓒTeemStone Pty. Ltd. All Right Reserved

� Solution

To solve the problem, onTune, which is the specialized solution for monitoring and analyzing system performance, was installed in the target system. And it found the root cause and solved the problem.

- Installed onTune in the Windows server including the problematical application

- Without certain configurations, the usage of CPU/Memory/IO for the entire system was saved by second unit, as well as the usage of CPU/Memory/IO for every single process with PID, and obviously those data was saved with every second interval monitoring.

- As a result of analyzing resource usage such as CPU or Memory of processes through onTune, the system operator found that the virtual memory usage of the problematical Flash program was steadily increased, and it showed down when it reached to the certain size.

- Based on this fact, the system operator requested the program developer to confirm the problem. Confirming the memory leak of the program, the developer took action for fixing the problem.

� lesson and learned

Thanks for onTune, system operators are able to implement logging resource usage of every single process by second interval, as well as the global resource usage of system. Furthermore, certain past time when problems happened is being monitored so that the users can analyze the resource usage of processes, so it must be helpful for figuring out the root cause of problems.

Pic1 : The whole CPU usage of the server in certain past time, and the every single process working on the server.