Slide decks : http://bit.ly/mayank-raj-serverless-cron
Talk Abstract
After a point every application relies on background process which is powered by cron. Instead of a having a cron server this talk shows how cron process can be designed as a micro-service with all the benefits that come with being serverless - scalable, manageable, maintainable and cost effective.

Talk Description
Instead of provisioning a dedicated server for cron jobs, we at the R&D team at Cactus Communications were looking for ways we can achieve higher time precision and get the job done at a fraction of cost. What we came up with was a way in which we use a few of the managed services provided by AWS (which have equivalents in all major cloud services) that gives us second level trigger ability, failure retries from the cron process and not the application, alerts, easy visual interface to see the current state and detailed logs right out of the box.

With this we did not introduce any new tool, technology or even package to pipeline. This meant there was no disruption to the workflow and developers did not have to look into a new tool or package. The switch was seamless. We are now alerted only when a predefined patterns are seen in the logs. As it uses managed service, we do not have to look into issues like scalability, availability and even disk usage, which have to be monitored in case of cron servers.