Aparna Chaudhary' Blog

I’m working on a jBPM prototype since last couple of days. The integration with Spring wasn’t a smooth ride. In my previous projects, I used jBPM3.2. But it seems jBPM4 is pretty much a rewrite. There are couple of few nice additions like service API’ for task and process management, support for java task and many more. And there are few changes which I didn’t like. Till 4.2 the support for JPA was provided, which is taken off starting 4.3 release. Yes definitely you can write some wrappers and still achieve it. But!! Between 4.3 and 4.4 also there are some major API changes. So the whole point is to get the app running, I had to digg into mailing lists and source code. Now that it works, I thought I would write a short blog about it.

Read more…

Recently I was playing around with Spring Integration. To understand any new framework one cannot just rely on documentation. So I created a demo application to try out different features of the framework. Since my main motivation was to understand Spring Integration framework, I wanted to spend minimal efforts in input data generation. So I decided to work with Twitter messages.

Most of you must be familiar with Twitter. Twitter is a service for friends, family, and co–workers to communicate and stay connected through the exchange of quick, frequent messages. There are different clients available to use Twitter. In the sample application, I first read the friends timeline. Then based on the source/client used for tweeting, the tweets are routed to different channels. The tweets originated from “web” are simply logged, while the tweets originated from “Dzone.com” are dispatched using a mail sender. The basic flow of the application is depicted in the following diagram.
Read more…

Spring source community is coming up with spring batch 2.0 in Q2-2009. Spring batch is the first java based framework for batch processing. I think the decades of experience of Accenture in enterprise batch processing really helped for defining the use cases.

Most of the batch applications need to process high volume business critical transactional data. While doing so some set of non functional requirements (NFR) are sort of mandatory in such applications. These NFRs include performance, scalability, restartability, repeatability. I worked with couple of investment banks and my experience says that such batch applications are developed based on either Messaging model or Multi-threading model. Lot of efforts and time is spent by architects, developers and testers in building this robust infrastructure for batch processing. Also we cannot overlook the cost involved. Whenever you move across projects, you end up creating your own batch processing framework. Sigh!!

Some nice features that are introduced in Spring Batch 2.0 are conditional step execution, finer metadata access control and chunk based processing. To perform chunk based processing, we need to configure the commit-interval in a step. The transaction is committed after number of items specified in commit-interval are processed.

<step id="step1" job-repository="jobRepository" transaction-manager="transactionManager">
      <tasklet reader="itemReader" writer="itemWriter" commit-interval="10"/>
</step>

Given the features provided and use cases handled by Spring batch, it can prove to be the de-facto framework for enterprise batch applications.