January 2019
Beginner to intermediate
154 pages
4h 31m
English
FIFO is the default scheduling policy provided by Spark. It is okay to use FIFO when you are in the learning phase, but it is not ideal for production applications. In the case of FIFO, when the jobs are submitted, their tasks execute in the order they arrive. This means, if you have a stage running on a Spark cluster that is occupying half of the resources, then another stage can execute its task on available resources. But, sometimes, this becomes a problem. If the first stage is a heavy one and consumes all the available resources, then the second stage will not be able to execute its tasks.
Read now
Unlock full access