Beam Bank Transaction Analytics

This repo contains small example for doing streaming analytics. The use case picked up is use case of money transfer event from an accountID to another accountID with a particular amount of money and an event timestamp.

---

There is two types of analytics windowing used:

Globally Aggregated Transaction

Analytics is done in global window. Here, global means unbounded window. So, all aggregated data created through this mode is considered as the sum of data from the beginning to endless time window.

Periodically Aggregated Transaction

This type of windowing can be configured into a certain time window. Which means, now we can calculate the aggregated value for a certain period of time, for instance per minute, per hour, per week, and so on.

Both mode mentioned above uses speculative aggregate trigger. This means in a designated period of time within a window, the application will trigger one or more speculative aggregates from the data which have arrived on to the system.

Getting Started

• Install maven (the newer version, the better)
• Run mvn test to execute the test.
• To run the application, chose either StreamingPipeline.java or BatchPipeline.java.

References

1. Apache Beam Game Example
2. Apache Beam Documentation