Thank you for your reply and thank you for very good comments. Let me address them one by one.
Makes, sense I need to use a more up to date architecture diagram.
2 & 4. The idea was to show what innovations Flink introduced and then to show that Spark is implementing similar features (e.g. Tungsten)
As far as I know Spark is still using micro batching, while Flink was using true streaming from the very beginning (more on this here: data-artisans.com/blog/high-throug...). Spark folks are also working on the continuous streaming feature but as far as I know it's know released yet.
Hi Jacek,
Thank you for your reply and thank you for very good comments. Let me address them one by one.
2 & 4. The idea was to show what innovations Flink introduced and then to show that Spark is implementing similar features (e.g. Tungsten)
As far as I know Spark is still using micro batching, while Flink was using true streaming from the very beginning (more on this here: data-artisans.com/blog/high-throug...). Spark folks are also working on the continuous streaming feature but as far as I know it's know released yet.
I was referring to this: ci.apache.org/projects/flink/flink...
That would be awesome! I would definitely visit a meetup like this.