Downloaded 30 times
Uploaded byDataStax Academy
Using Event-Driven Architectures with Cassandra
The document discusses the implementation of event-driven architectures (EDA) using Cassandra, highlighting its real-time capabilities for business operations. Key concepts such as events, emitters, and sinks are introduced, along with the benefits of extreme decoupling and asynchronous processing. It also covers various EDA styles and provides practical tips for effective integration with Cassandra.
Using Event-Driven Architectures with Cassandra
- 1. @petabridge Petabridge.com Event-Driven Architectureswith Cassandra Aaron Stannard, CTO & Cofounder Petabridge DataStax MVP 2015
- 2.
- 3.
- 4.
- 5. @petabridge Petabridge.com A RadicalNew Product
- 6. @petabridge Petabridge.com Allow ourcustomers to start conversations with specific types of users.
- 7. @petabridge Petabridge.com Had tobe done in real-time for best results.
- 8.
- 9.
- 10.
- 11.
- 12.
- 13. @petabridge Petabridge.com I hadSTATEFUL, REACTIVE, STREAM PROCESSING problem!
- 14.
- 15. @petabridge Petabridge.com Why DoWe Care About EDA?
- 16.
- 17.
- 18. @petabridge Petabridge.com Key Terms •Event – a significant change in “state” • Message – a notification of an event • Emitter – detect, gather, transfer events • Sinks – react to events immediately • Channels – conduit between emitters and sinks
- 19. @petabridge Petabridge.com Goals &Benefits • Extreme decoupling • Easily distributed • Inherently asynchronous and concurrent
- 20. @petabridge Petabridge.com EDA Styles •Simple Event Processing • Event Stream Processing • Complex Event Processing (CEP)
- 21. @petabridge Petabridge.com Event StreamProcessing w/ FSMs
- 22.
- 23.
- 24.
- 25. @petabridge Petabridge.com CQL Schema CREATETABLE IF NOT EXISTS FSMState ( persistence_id text, state_id int, state_data blob, PRIMARY KEY (persistence_id, state_id) )
- 26.
- 27.
- 28.
- 29. @petabridge Petabridge.com EDA +C* Tips and Tricks • Large number of small sinks works best • Define simple, reusable journals and snapshots • C* == durable backup, App = single source of truth • TTL everywhere your business domain supports it
- 30. @petabridge Petabridge.com Have questions?Ask us! http://petabridge.com/
Editor's Notes
- #17 C* users should care about their application architecture for the simple reason that C* isn’t magic and can’t do everything. Cassandra can’t solve the problem of real-time correlation or reaction at the application layer. Developers have to solve these problems themselves, and Cassandra is just one part of the solution.
- #21 Event Stream Processing and CEP play nicely with Cassandra, and that’s what we’ll be focusing on
- #26 Intentionally small and stupid