News for hackers. Stuff that matters.
Re:A scikit-learn pipeline in Wallaroo by nitbix on Thursday February 08, 2018 @07:50AM (#16331752)
Hi everyone! I'm the author of this post, if you have any questions please post them here, on our mailing list or on IRC, and I'll be glad to help!
Re:A scikit-learn pipeline in Wallaroo by dominotw on Thursday February 08, 2018 @11:35AM (#16333359)
I am trying to understand where Wallaroo fits in the steaming landscape.

How does this compare to kafka-streams app running on kubernetes

Re:A scikit-learn pipeline in Wallaroo by spooneybarger on Thursday February 08, 2018 @11:57AM (#16333540)
Hi,

VP of Engineering at Wallaroo Labs.

The folks who we most appeal to right now are folks who are looking for a native Python streaming solution. There's also a decent amount of excitement from the Go community about our Go version that is out as a preview release.

There's a lot of JVM based solutions out there for streaming, the choices outside the JVM are pretty sparse.

Wallaroo is particularly well suited to streaming applications that need to keep some state in-memory to make fast decisions based on incoming data.

The Python version of Wallaroo can handle tens of thousands of events a second and respond within a few milliseconds. This is all dependent on the application logic of course, but it should give a rough idea.