We have many development teams here at Principal that would also be happy if compute could scale up and down via scheduling as we have hot times that occur from time to time and other times would like to downscale to lower compute. Thanks!

We would also like to see the idea of creating auto-scale schedules to accommodate our weekday/weekend application workloads similar to what ParkMyCloud offers for powering on/off cloud based servers.

This is a great suggestion. CPU is seldom the reason for scaling, neither is memory. Often we know exactly when systems are busy on a daily basis, based on Corporate customer consumption. We would like to scale up certain times of the day. And in most countries, (for business apps) not on weekends.

A variant of this idea could be to have a "weekend" mode which auto-scales based on previous weekend workloads.

We want something like this too! We currently use auto scale now, but since it takes too much time to scale up (after 1 hours CPU load only, and yes there is another idea for changing that: https://feedback.mongodb.com/forums/924145-atlas/suggestions/40797049-autoscaling-cpu-memory-utilization-duration-mon) we want to scale before the users are gonna actual use our application. This idea was suggested by the support desk after I ask questions about this. It would be great we can made a sort of schedule about how to scale. For example: Saturday @ 22.00 hour, scale down tot M30 Sunday No changes Monday @ 10.hour scale up to M40 Thursday @ 08.00 hour scale up to M50 Thursday @ 18.00 hour scale down to M40 Wednesday @ 08.00 hour scale up to M50 And so on...

We are currently solving this with AWS lambda scripts, calling API's , but it would indeed be great that there was some sort of scheduler for up and down scaling.

Hi Anand, This certainly makes sense and is something we'd like to explore in the future. Out of curiosity, have you looked at programmatically driving this on a cron job with the Atlas API or Terraform provider? Cheers -Andrew