Apache Spark Integration and Platform Execution for ML - ML 073
Adventures in Machine Learning
00:00
Datarix Spark
A lot of the Hadoop and MapReduce operations worked on disk. Spark completely transformed all those operations to memory. Reading one megabyte sequentially from memory is a hundred times faster than reading it one megabytes sequentially from this. In modern day cloud based implementations like Datarix, Datarix Spark is not going to have that same relationship.
Transcript
Play full episode
