AWS Instance Scheduling - Normal Online Schedule
Hi folks,
I, on behalf of the central team (BEI), noticed many of you quite interesting in applying AWS Instance (EC2 and RDS) scheduling, especially for non-prod environments. I reached out to our CTOs discussing potential impact and risk of having this type of initiatives in our organisation.
There are important highlights resulting from the discussion:
- We agree that the normal online schedule for non-prod environment is 08:00 to 22:00 (GMT +7) in weekdays only.
- If you follow normal online schedule, that will be super beneficial for non-critical and non-platform services, then you will not need to notify your dependants services and QA/SDET team.
- If you want to have more strict schedule (e.g. 09:00 to 19:00) for your non-critical and non-platform services, then you will need to notify your dependent services and QA/SDET team.
- Potential cost saving, assuming all services (including platform and critical one) join this initiative, is about $19,000.
With this post, I want to create a discussion with all you about:
- How should we schedule our platform and critical services? Our proposal
- Keep it running 24/7 in staging to minimise any impact on hotfix and non-office hours delivery problems.
- Use the same normal schedule with other services to save more, but prepare a easy turn on script specifically for these services.
I will create a thread about this in #cost-optimization-2020 and be sure only use this thread to discuss about this topic. It will make it easier for anyone that want to contribute to jump in.
In same time, to track this initiative progress and interest, we will use this sheet to make sure everyone can find out other product domains status and potential schedule conflict. Every product domain should add their plan toward their instances on each AWS account they used.
Together we can 
:muscle::skin-tone-3: