JBatch or not such a big data
A story about what kinds of problems encounter in design of modern systems.
Offline data processing still goes on in almost every Java-project for large companies. But there are not so much data, and passing them through the Hadoop would be shooting at sparrows with cannons. In most cases such processing of "not so big data" is performed by small programs, or, at best, in-house applications. They are often ideally suited for a specific task, but very capricious in case if something needs to be changed/adapted/expanded. All this ends with the complete rewriting. Scalability of in-house solutions often suffers.
To make clear this important but little underrated aspect, JSR-352 Java Batch was developed. During the session listeners will become familiar with the JSR. They will learn how to define "work", "objectives", as work is broken down into "steps" and "pieces". Dmitry will consider how to parallelize tasks and ensure their transactionality within the container. And most importantly, how to not repeat. Also JBatch will be compared with Spring batch.
This framework has been officially included in the Java EE 7, so that it is available "out of the box" on any Java EE 7 server.