Event Streams question: ESP Platforms

[Yury Nebieridze]

What are the best ESP Platforms for particular Event Stream processing scenarios and why?

[Alexander Dean]

The hottest ones at the moment are probably:

* Kafka Connect
* Flink
* Spark Structured Streaming

There's also a more "descontructed" approach of chaining together serverless functions like AWS Lambda.

[Ron McLeod]

Are there any implementations which are suitable for low CPU/RAM platforms?  I've put together some applications using Kafka/ZooKeeper which do what I want, but were pretty demanding on RAM.

[Yury Nebieridze]

Thanks for the answer, Alexander.
You mentioned "There's also a more "descontructed" approach of chaining together serverless functions like AWS Lambda." .

Will it be a multi-phase analytics system? Meaning a mixed strategy of analytics processing, chaining:
- at-the edge analytics (IoT devices)
- in-stream analytics (derived/complex stream evens)
- at-rest analytics (rich, historical context in data lakes for predictive analytics modeling and forecasting, data patterns discovery)  

Can you elaborate on that? Any specific scenarios/patterns/use cases for building such chains, which, I understand should mix both push-based continuous intelligence analytics systems with on-demand, pull-based analytics.