Storm micro talk
2 likes878 views
This document summarizes Storm, an open-source distributed real-time computation system. It discusses how Storm can be used for real-time processing of data streams from logs, social media, sensors and advertisements. Storm provides topologies of spouts that emit streams and bolts that perform computations, with at least-once processing semantics. It also offers fault tolerance and horizontal scalability similar to Hadoop.
Storm micro talk
- 1. Storm micro-talk Big Data Beers #4 10/04/13 Pere Ferrera Bertran
- 2. Talking about Real-time processing Stream processing Real-time stream processing ... <Insert-your-favourite-BuzzWord-here>
- 3. Examples Logs Social media Outlier detection / trends / tracking / ... Sensors Quality-Of-Service / alerts / ... Ad impressions ...
- 4. Options 1. Queue / Workers 2. Actors Akka, Erlang, Scala, ... 3. 1 + 2 4. ... Nathan Marz thought things could be better.
- 5. Storm Higher-level paradigm Topologies: Spouts + Bolts
- 6. Storm Groupings At-least-once semantics timeout ack(), fail()
- 7. Storm A-la-Hadoop: Fault-tolerancy Horizontal scalability
- 8. Trident hackaton "Cascading for Storm" Higher-level than Spouts and Bolts Disclaimer: I'm no Storm expert! Disclaimer: Drinking more beers won't make it any better! Have fun!