Managing infrastructure that runs millions of programs in Turing-complete languages is a lot easier when execution time is known to be X, rather than potentially being infinite.
It also transfers a lot of complexity and responsibility for Lambda from AWS engineers onto the users of Lambda.
This setup makes sense to me, and I don’t see a need for it to change.
If you have really long-running processes, a different AWS service like Elastic Beanstalk (or your own self-hosted Lambda-like service) is a better fit.
Parse.com did the same thing, as did all of those old-school PHP hosts, so it's been a common solution for at least the 22 years I've been using hosting services -- maybe longer.
4 comments
[ 2.4 ms ] story [ 17.6 ms ] threadIt also transfers a lot of complexity and responsibility for Lambda from AWS engineers onto the users of Lambda.
This setup makes sense to me, and I don’t see a need for it to change.
If you have really long-running processes, a different AWS service like Elastic Beanstalk (or your own self-hosted Lambda-like service) is a better fit.