Maintaining on-demand servers

You could use the AWS Instance Scheduler to orchestrate the scheduling. You can configure schedules appropriate for each of your workloads in the DynamoDB table that controls the Instance Scheduler, and you can associate each instance with the appropriate schedule with a tag you configure on the instance. You can expose the tag in the EC2 instance view in the console to see easily which schedule is assigned to each instance.

You would need to configure the schedules in Instance Scheduler to include the patching window appropriate for each workload or patching code, depending on how you classify your workloads.

The Instance Scheduler is documented here: https://docs.aws.amazon.com/solutions/latest/instance-scheduler-on-aws/solution-overview.html. The configuration of schedules is explained on this page in the documentation: https://docs.aws.amazon.com/solutions/latest/instance-scheduler-on-aws/schedule-reference.html and tagging instances with schedules is explained here: https://docs.aws.amazon.com/solutions/latest/instance-scheduler-on-aws/operator-guide.html#tag-instances-for-scheduling

Note that if Windows servers are included in your patching process, it often completes rather quickly but sometimes takes substantially longer. If you trigger a power-off via EC2, whether it's with a Lambda function, through the console, or with the Instance Scheduler, EC2 will only wait for some minutes for the instance to finish powering off, and if it doesn't, EC2 will power it off anyway. As you may know, that may corrupt the whole Windows installation. The only safe way to power a Windows server off is to trigger the shutdown from inside Windows, so that the operating system is in control of the power-off. The AWS Instance Scheduler or your custom Lambda would only be responsible for powering the instances on. The Instance Scheduler also supports "start only" types of schedules.