Work with EMR for machine learning

- [Narrator] So the next service we're going to look at…is Elastic MapReduce, which is managed Hadoop,…Spark, and other type of library clusters…of virtual machines.…So, the question is, why should we use virtual servers…when we have API's, docker containers,…and many other options?…Well, the answer is, you shouldn't always.…But there are situations for which…you need the level of control.…Could be security requirements.…Could be custom setup steps.…

Could be the amount of data.…I've been working with some bioinformatics customers,…and in processing genomic sequencing result output,…the data is huge, and taking advantage…of the economies of spot pricing on EC2 is really critical…for some machine learning workloads.…Amazon Elastic MapReduce is platform as a service.…It's Hadoop clusters, so master and worker nodes,…that are customized EC2 instances,…that are designed to run Hadoop…and its associated libraries,…such as Spark, SparkML, or machine learning,…and other workloads.…

Many data processing libraries such as Spark and SparkML,…