We are using PlayFab gameserver features with the t2 nano servers and unity based game servers. GameServer is very very lightweight with message rate of around 10-15 per minute and each server has only 2 players. FrameRate of the server is limited to 60 because it will go on the roof if we don't set it.
We have a specific issue on the first game process on each t2 instance. We easily run 4 processes per AWS instance and I guess we can even increase it to larger values. The issue is that when first process is created, the game client either would not be able to connect to it or the connection will take 10-15 seconds to happen.
We put servers in Japan to be far from both US and Europe to find latency related issues sooner rather than later.
Until yesterday we actually were having difficulty connecting to first processes on instances most of the times but after I i made our server startups more lightweight, now I can most of the times connect to it but too late.
My question is, What is exactly going on when first instance is getting started. Do you also launch your own monitoring software and other stuff lazily when the first process starts?
Can I know a bit more about it and can we improve the situation? How bigger servers behave when we go to prod in this regard?
I imagine we would get bigger servers and run more processes on them for prod if we find out that our game performs better and more efficiently on bigger servers when testing.