This repository has been archived by the owner on Jul 22, 2018. It is now read-only.
Auto scaling
4 speakers
- Why AWS? hurricane season, capacity constrained... moved things that they could to AWS
- Autoscale because of cyclical pattern of usage
- And big events like Hurricane Sandy
- exceed 80% utilization, add 10% cluster size
- drop below 65%, go down incrementally
- timeboxed
- monitoring tools suck, especially in a dynamic environment
- rolling out app updates
- zone failures can lead to more zone failures if the problem's on your side (because of increased traffic? unclear)
- non critical apps with static autoscaling
- instance failure -> 15 mins -> recovered
- cloudinnit, puppet tree built into an RPM, pushed to S3 as yum repo
- stricter uptime needs
