- set up lambda (basic execution role)
- just api call and return
- incorporate boto3 writing to dynamodb
- change input to handle date range or payload etc
- change success message
testing in browser - woo fun, but let's make an api endpoint
- make api endpoint with POST
- test in browser
schedule that bad boy
- configure cron
great, we have a job. what happens if it fails?
- intro to cloudwatch logs
- configure SNS etc
- demo email or sms notifications
okay cool but no one uses stdlib and i need to write to a db homie
- same thing, with requests and db module
- run deploy for pushing lambda, all other things equal
test curl from CLI
- api.databae.io/cats
- free up to 1M requests (free ETLs - can't say same of spark)
- lambdas run in response to stuff, s3 uploads, SNS notifications, etc.
- gotcha - iam roles might be more restricted - YMMV, but SG/permissions are super easy regardless
- limits (5 min, 1.5 GB) can be overcome through chaining etc, or you can go docker/stored procs/etc
- hope this blows your mind like it did mine