Friday, March 8, 2013

Liferay search container with curd operations.


Liferay search container with curd operations.

I create port let which explained how to use life ray search container to display records as grid.
I added some of records in person tables and I display all records in liferay grid and when click on any row it will navigate to update form, once we finished upadate it will navigate to search container page.
The following the link to get search container portlet.

Steps to run the portlet:
1.      Download portlet
2.      Place the portlt into liferau plugins/portlet directory / if you are use Liferay IDE create liferay project from existing source
3.      Run ant build-service
4.      Run ant deploy
5.      See the sample category and drag and drop the port let
6.      If records available you can see the records otherwise add new person by click on Add new person link.
7.      When you click on liferay grid row you can update the person record.

Screen Shots:


Note: This is done in liferay6.0.6 if you are using liferay 6.1 then create your portlet and manually copy file to your portlet. When you copy portet.xml , liferay-portelt.xm. you should not copy directly because dtd version changed so make sure dtd version should acceding to liferay versions.

Friday, March 1, 2013

Connection pooling in multithreaded applications


Connection pooling in multithreaded applications

Introduction
This article talks about developing design approaches for multithreaded applications performing database operations. While developing a multithreaded database application, we always end up scratching our heads answering (or trying to answer) mischievous questions as:
  1. Who should create the connection object?
  2. Should a connection object be a property of the business object? Or should it be used as a utility?
  3. How to handle transaction operations?
  4. Who should dispose the connection object?
Looks like seven W's of Wisdoms are making enough noise in our head. I found my way of answering these questions, and I am sharing them here with you. I have tried two approaches for it, which are described here.
Single connection object, multiple transactions objects
I have used the Singleton design pattern to make sure that only one connection object is created. The same connection object is shared across multiple threads. It is now every individual thread's responsibility to handle the transaction. Every thread will create its own transaction object and will pass it to all commands it will be executing. So straight away, the thread should maintain its own commit and rollback policy. Thus, a single connection may execute multiple transactions simultaneously, wholly managed by the calling threads.
As parallel transactions are not supported by the database, make sure the code block handling the opening of the connection and transaction creation is executed under a lock mechanism. Thus, in this approach, though we can use the same connection object, we have to make sure the code is thread safe (use of lock).
Advantages:
  1. Implements a Singleton, shares one connection object across multiple calling threads.
  2. No need to dispose the connection object (but you must call the close method).
Disadvantages:
  1. Does not use the connection pool feature, as only one connection object is created and used.
  2. Increases execution time as the command must be executed using the same connection object.
Here is the block diagram for the singleton approach:
Multiple connection objects
The multiple connection approach is slightly different. This one gives the calling code control of the connection object. It becomes the calling code's responsibility to use and dispose the connection object. In an ideal scenario, the calling code (Business Layer) should not hold the reference of the connection object. It should request for the connection object when required, and use and pass it to its sub-routines, if required. Thus, the connection object is opened in a method and closed in the same one. The pseudo-code will look like:
Method (){
 - Create local connection object, open it
 - Do transaction with database.
 - Pass connection object to other methods called from this.
 - Commit or rollback transaction
 - Close connection, dispose it
}
This approach allows us to create multiple connection objects and execute them in parallel. But, it also enforces some conditions as:
  1. Calling code should take ownership of the connection object.
  2. Calling code should handle the connection pool by declaring it in the connection string.
  3. As there could be multiple connections opened simultaneously, it's the calling code which must maintain the relationship between the connection and its transaction.
Advantages:
  1. Uses connection pool to create multiple connections simultaneously.
  2. Faster compared to singleton, as multiple connections will execute their own transactions.
Disadvantages:
1.       Need to make sure that the connection object is disposed properly.
2.       Connection object needs to pass through methods.
Here is the block diagram for the multiple connection approach:

Recent Posts

Recent Posts Widget

Popular Posts