​
Execution Time

Any one of our Resources will have a time to start and to end. These are affected by two things:
  • Timezone, decided at Resource creation
  • Tolerance, in time, for the action to be taken
Tolerance is what our current cloud systems use to distribute our workload, in minutes; so a Scheduler set to start at 00:00 usually does at 00:01 ~ 00:04.
This is known as the “Thundering Herd” problem, and it is a consideration for every Cloud provider (AWS, Azure, GCP, …)
Also note, there’s (currently) no SLA guaranteeing execution.Our service is very reliable, but we don’t recommend using Schedulers with no allowed tolerance for mission-critical systems.
These tolerance not only help with ensuring the action will definetely take place, but often Cloud providers need a lot of time to start, stop or modify their instances anyway. Usually, snapshots are taken and stored, database backups take place, diff logs are transfered between replicas. Until changes are actually in effect, resources are usually charged at their previous levels. One of the reasons we recommend turning on Account Logging is for you to be able to check on them and see their day to day timings. It will be registered when the action took place and when their status was actually confirmed to have changed.

​
Editing

Schedules & Adjusters are create-only; if you need to change a planned schedule, delete the Resource and create them again.

​
Planned Features

Regarding execution, we currently are working in two features. They are:
  • Precise timing: we’ll strive to start execution at the planned time, within seconds.
  • Alerts: in case of a long error about starting or stopping a Resource, with significant cost involved, we’ll send alerts to you over selected channels.