incubator-livy VS Apache Hadoop

Apache Livy - REST API For submitting jobs to a cluster. Used in conjunction with Jupyter or Zeppelin notebooks and you have a multi-tenant SQL only workload — https://livy.apache.org

There’s definitely some patching that still has to happen depending on the version (just found this Spark 3.2 PR), but I presume many folks that actively use it have to maintain their own forks. Latest version of EMR is Spark 3.1.2, but 3.2 is coming soon.

Apache Livy - an open-source REST API for interacting with Apache Spark from anywhere.

Spark 3 has some solid support for Kubernetes. Livy is a good addition to Spark on K8S, although support is not merged upstream yet (there's publicly available images and other in the PR). I'd strongly recommend to avoid running HDFS on top of Kubernetes. Either cloud-native buckets or on-premise bucket storage like Minio would be much better suited.

Did you check out tools like https://hadoop.apache.org/ ?

check out the various dockerfiles bundled with hadoop on GitHub. you can point to them from within docker-compose. they haven't been updated in a couple years tho.

There are different ways to implement parallel dataflows, such as using parallel data processing frameworks like Apache Hadoop, Apache Spark, and Apache Flink, or using cloud-based services like Amazon EMR and Google Cloud Dataflow. It is also possible to use parallel dataflow frameworks to handle big data and distributed computing, like Apache Nifi and Apache Kafka.

There are several frameworks available for batch processing, such as Hadoop, Apache Storm, and DataTorrent RTS.

A copy of Hadoop installed on each of these machines. You can download Hadoop from the Apache website, or you can use a distribution like Cloudera or Hortonworks.

The Apache™ Hadoop™ project develops open-source software for reliable, scalable, distributed computing.

So, clearly with your AP CS class and PLC logic knowledge, if you were dumped into a codebase like Hadoop, QT, or TensorFlow you'd be able to quickly and competently analyze what is going on with that code, understand all the libraries used, know the reasons why certain compromises were made, and be able to make suggestions on how to restructure the code in a different way? Because I've been programming for coming up on two decades and unless a system is within the domains that I have experience in, I would not be able to provide any useful information without a massive onboarding timeline, and definitely wouldn't be able to help redesign anything until actually coding within the system for a significant amount of time.

This requires the use of distributed computation tools such as Spark and Hadoop, Flink and Kafka are used. But for occasional experimentation, Pandas, Geopandas and Dask are some of the commonly used tools.