Running pipelines on AWS?

[rsignell-usgs]

Hello! Newbie here. :)
Any tips for someone who would like to try an xarray-beam rechunking pipeline on AWS?
(USGS is wedded to AWS at the moment)

[alxmrs]

Sorry for the late reply. I'm not too familiar with how we can run this on AWS (I haven't done it myself), but here's how I would do it:

• Set up a (py)Spark cluster in AWS EMR, see the AWS docs or this step-by-step Medium post.
• Specify the SparkRunner in the --runners flag of your python Beam Pipeline.
  • See also: https://beam.apache.org/documentation/runners/spark/#running-on-a-pre-deployed-spark-cluster
  • Likely, you'll need to install apache_beam with AWS's extra requirements: pip install apache_beam[aws].

I hope that helps.