This is what I wondered about for a while - in particular, how to turn on the container or server after receiving a connection request (getting it to sleep after x minutes of inactivity is pretty easy).
Say I had an EC2 instance on AWS that was turned off. After going to its DNS URL or IP address, I would like the server to wake up and serve its contents for the client. Heroku is currently dealing with this, and I have always wondered what happens behind the scenes ... and I know that cloud servers like AWS usually don't have wake-on-lan functionality.
I had some ideas about what could happen - perhaps the sleeping dinosaurs are not actually disconnected, they just sit idle. It is also possible that they are containers attached to a larger “parent” server, which receives a connection request, spooling sleeping servers and forwarding connections to the activated server. Does anyone with architectural knowledge know what secret sauce is?
As you can probably say, I know almost nothing about dev ops or management servers. This is just what scratched my curiosity.
source share