Hi Alexandre, I’m not aware of any effort to port the Flink API to Storm. The closest thing to what you are looking for in Storm is a new “Streams” API that closely resembles the Java 8 Streams API, allows usage of lambdas, etc. That API will likely become available in the upcoming Storm 2.0 release.
A word of caution regarding Flink’s Storm compatibility layer, there is some important fine print [1] to be aware of. I’m also unsure of the future of that effort as it hasn’t been appreciably updated in almost 2 years aside from a recent update to package names for compatibility with Storm 1.0. -Taylor [1] https://github.com/apache/flink/blob/master/flink-contrib/flink-storm/README.md > On May 19, 2017, at 3:29 PM, Alexandre Vermeerbergen > <avermeerber...@gmail.com> wrote: > > Hello, > > Those days, it's becoming difficult to choose betwen Storm and Flink. I love > Storm bolts & spouts philosophy, but I'm missing the even-time based sliding > windows promised by Flink. > > Flink offers a solution to run Storm topologies. > > It's nice, but we're quite used at setting up Storm clusters, with > scalabilty, HA and upgrades in mind. Redoing everything with Mesos or one of > the other scheduler supported by Flink is feasible, but it will require some > time for us to be as production-ready as we are today with Storm. > > Since we're so much used used to Storm, we'd love to see the opposite option: > ability to use Flink high-level API and semantics over Storm. > > Is there any such project that'll eventually allow running Flink code in a > Storm topology? > > Best regards, > Alexandre Vermeerbergen.
