A few bullet points on Introduction to AWS Glue (Ian Robinson - Live from the London Loft)



  • A few interesting things, but not worth the hour.
  • Slides have a ton of words on them that are hard to read since he doesn't really slow down for them. (Sometimes hits the main points on them, but it feels like sometimes he doesn't get to everything)
  • Does a few copy/paste without really explaining what's going on.
  • Talks a lot about things Glue can do, but doesn't show many examples.
  • Examples are spread out. I'd prefer doing something from start to finish.
  • Also, the example is relatively complex. Part of it creates an internal lambda function. That's not very well explained.
  • One good thing is pointing out the hierarchy where you name stuff in S3 like year=2019/month=6/day=9/some_file.csv and it automatically picks up the date as something you can querty against.
  • Talks about Notebooks and Developer endpoints and how that makes it a lot easier to develop, but doesn't show it. Would be great if he had time for that.
  • The sample he does at the end takes several minutes to run. Would be better if the sample data was trimmed down to run quickly for purposes of the demo.
  • The part at the end (44:15) talking about bookmarks is helpful. Knowing that by default, Glue won't process the same S3 file twice.