Skip to main content

Improving the ADF Application Performance

Following are some of the important points to increase the performance of your ADF Application.

1. In view object tuning section always set the fetchSize to value which will be greater than the no of rows you are showing on the grid
Because


Default fetchSize is 1 with this fetch size for each row it goes to DB and fetch the records, that is one by one as shown in above diagram.
I you want to increase the performance set some value for this fetchSize attribute so that it will fetch the rows in batches.


     If you are showing 10 rows in a grid then in that ViewObject set fetchSize as 11 (row needed+1).  FetchSize is nothing but how many rows have to be fetched from DB in one trip.Using this no of Db trips will be reduced.

2. Remove unwanted bindings from pagedef and refresh conditions on iterator(only keep the required value of this property).

3. Activation and Passivation :
    - when you run your ADF application first time,bydefault ADF creates PS_TXN table in the schema to store the details

  - Do not passivate the state of a transient attribute in ViewObject. If you include all transient attribute from tuning section then all the calculated and transient attributes will be passivated and when the session state is reloaded they are activated .
  




 
Best practice is,if you need specific attribute then you can do this by using the attribute property  editor and select the passivate check box.





5.Set partialSubmit="true"  wherever it is possible on buttons,links,menu items.It will refresh only part of the page and not the whole page and increases the performance.

6.In ViewObject use the Access mode as RangePaging instead of Scrollable.



 
  The main difference is if you use scrollable it keeps all the rows fetched from the DB in to the memory as you scroll down but in case of RangePaging it stores only those rows in memory which we are seeing on the screen and it will avoid the heavy uses of memory for Data.
  




7. Use small memory scopes for beans.

8. Make the ids of components as small as possible.
   for folowing component the ids should be of max 2 characters.
  
   <af:pageTemplate>
   <af:region>
   <af:panelCollection>
   <af:table>
   <af:treetable>
   <af:tree>
   <af:iterator>
 
  Because these are container components and Child components use their parent component id.
  This is good for browser performance.

9. Use the following steps to optimize your application performance
  In BC ->
  - Detect and tune the VO's slow queries.
  - Setting the appropriate tuning-values on the VO.
  - Use VO RangePaging if the table rows > 300.
  - Use lazy loading on AM's
  - Sizing the application module pool.

10. Set the proper JVM heap size and choose an effective garbage collection strategy.

Comments

Popular posts from this blog

Passivation and Activation in ADF (Application Module )

1. For performance reasons, ADF keeps a pool of application modules in memory. It tries to give each session the same application module as the session used during the last request; however, this might not be possible during peak load of your application. 2. In this case, ADF saves the application modules state in a database table so the application module can be used by another session. This is called passivation . 3. When the first session needs the application module again, its state is retrieved from the database process known as activation . 4. If you have made an error in your code and depend on some variable that is not persisted correctly when your application module state is stored, you will experience mysterious errors under high load.   Enable/Disable Application Module Pooling : Right-click on your application module, choose Configurations.By default, each application module has two configurations. Ensure that the one ending in …Local is selected and then click

Get modified rows from Entitiy Cache

To get the modified rows from entity cache we have getEntityState() method at EntityImpl class. Refer to my previous blog  Accessing EO impl methods from VO impl  where i am overriding the getEntityState() in EOimpl and calling it in VOImpl. We can use methods written or overridden in VOImpl class to AMImpl class. There are different states associated with an entity object. STATUS_UNMODIFIED STATUS_MODIFIED STATUS_NEW STATUS_DELETED STATUS_DEAD We have to check the state or row in our AmImpl class by using the VOImpl method and through this we can distinguish the rows present at vo. Add below code in AMImpl class along with my previous post. public void geCachedRowsCount(){         JobsVOImpl jobsVo = (JobsVOImpl)this.getJobsVO();         RowSetIterator iter = jobsVo.createRowSetIterator(null);             while(iter.hasNext()){             Row row = iter.next();             byte state = jobsVo.getEntityState(row);             System.out.println("Job_id -&

The file store "WLS_DIAGNOSTICS" could not be opened

WLS_DIAGNOSTIC ERROR weblogic.store.PersistentStoreException: [Store:280073]The file store "WLS_DIAGNOSTICS" could not be opened because it contained a file with the invalid version 1. A file of version 2 was expected. When you get this error while running your application on internal weblogic server delete the following file WLS_DIAGNOSTICS000000.DAT search the file in following path C:\jdev_work\system11.1.1.5.37.60.13\DefaultDomain this file is in DefaultDomain folder of your jdev. and delete the WLS_DIAGNOSTICS000000.DAT file . and run your applicatuon