1. The framework sample that you use, keep close eye on on following:
a. Every library included in project.
b. Every xml created that's load on runtime.
c. Every default code generate by tool
all above items will decrease a chances of bad performance for application.
Review a debug logs and try to resolve any unknown error and exception that caused but not disturbing application process but it might overloading JVM or database and other related applications / connections.
You might not able to feel any performance issue on development / test server because application is running with sample data / limit number of records and users.
2. Not every architect makes a good decisions:-
Custom measuring for example:-
a. Was impacted by garbage collection
b. just measured overall time but not # SQL Exceptions
DON'T ASSUME you know what code does somebody else developed
3. Use performance test tools dynatrace, New Relic, AppDynamics, HP Load Runner.
4. Idenfity how many sql exection in your code also number of same sql statement executed in your code.
a. Optimize the sql statement or cache data
b Connection equitation time
5. Identify how many number of threads are created and utilized by application and number of call made to API's and time take by each in API.
6. Bad Deployments
a. Different framework / library version between development and production server.
b. Perform Unit Test, Functional Test, Performance Test before deployment
7. Push without a plan ( Idea was great but implementation wasn't correct)
a. Need to check number of images on each page that might cause application slowness.
b. Number of redirection and size of resources are used in each page.
c. Use diagnostics tools to make sure overloading won't happen.
a. Every library included in project.
b. Every xml created that's load on runtime.
c. Every default code generate by tool
all above items will decrease a chances of bad performance for application.
Review a debug logs and try to resolve any unknown error and exception that caused but not disturbing application process but it might overloading JVM or database and other related applications / connections.
You might not able to feel any performance issue on development / test server because application is running with sample data / limit number of records and users.
2. Not every architect makes a good decisions:-
Custom measuring for example:-
a. Was impacted by garbage collection
b. just measured overall time but not # SQL Exceptions
DON'T ASSUME you know what code does somebody else developed
3. Use performance test tools dynatrace, New Relic, AppDynamics, HP Load Runner.
4. Idenfity how many sql exection in your code also number of same sql statement executed in your code.
a. Optimize the sql statement or cache data
b Connection equitation time
5. Identify how many number of threads are created and utilized by application and number of call made to API's and time take by each in API.
6. Bad Deployments
a. Different framework / library version between development and production server.
b. Perform Unit Test, Functional Test, Performance Test before deployment
7. Push without a plan ( Idea was great but implementation wasn't correct)
a. Need to check number of images on each page that might cause application slowness.
b. Number of redirection and size of resources are used in each page.
c. Use diagnostics tools to make sure overloading won't happen.
No comments:
Post a Comment