- Fast, in-memory (100x faster) or disk (2-10x faster). See Daytona GraySort contest and Official Result.
- Usability: rich APIs (Scala, Java, Python), concise, interactive shell
LoC of Spark in comparison with other projects
LoC of Spark core-framework and its integrated libraries
- Well designed, unified: Spark is a general platform and SparkSQL, SparkStreaming, GraphX, MLib are standard libraries included with Spark. These libraries provide a wide range of features that support multiple usages.
- Concrete foundation: Databricks & UC Berkeley AMPLab & Community
- Many adopters: Amazon, Yahoo!, Autodesk, Technicolor, Baidu, Celtra, eBay Inc., IBM Almaden, SamSung SDS, Sonny, …