what makes Flink better?

[blatoo]

Hi all,

I am new in big data world. Last Week I heard a Presentation about Flink. But I am still a little confused about Flink.

1. What makes Flink better than Hadoop and Spark?
2. Hadoop follows the MapReduce Paradigm, what kind of Paradigm does Flink follow?
3. I heard about a lot of advantages from Flink, but I am still want to know some problems in Flink, which part is still not so good compare to Hadoop and need to be improved in the future.

Thanks very much in advance!

Conny