unique pivot
Hello, I want to make sure why Multiplayer Server / Dynamic Standby works differently from my expect.
Build Setup
Servers per VM: 25
Target standby: 25
Maxumum: 100
Dynamic Standby Setting
25% x2
15% x3
Now
Current Active: 6 (And Active never exceed 10)
Since (25-6)/25=76%>50% I expected that the current standby is 25, but it keeps 50 now. What the reason it start a new one?
gilded sky
If i understood correctly your question,its due to the vm config you are using.
Lets do this: Active-hot, Standby-warm, not shown-cold. Server meaning the game. VM the machine in which your game is running on.
A standby server is warm meaning it is ready to be used for your game. you can instantly connect to this server. Active means it is in use by players .
When you set a standby count, you are telling playfab the number of servers which should be ready to be called upon at any cost. A machine takes some time to be ready for use so it is cold.
Now lets say you are using a vm type with 4 servers on it.
You set the standby count to be 4.
When the standby count falls to 3 lets say, playfab will start the process of giving you another machine with 4 servers on it. Why 4? As you need 4 servers to be standby and as the vm config you are using is for 4, it cant just give you 1 server and call it a day. It has to give you 1 whole vm instead of dividing it up as youu cant just cut a vm in half.
Now regarding dynamic standby. It is not to reduce the count of standby but to scale up when the demand of the game reaches x percentage.
Lets say you had 8 servers on standby where each vm supports 4 servers. Players start to join in a short period of time and now you have only 3 servers free left. Dyanmic standby would allow you to rapdily scale up the standby count. So instead of only deploying 2 new vm as 3+4<8 , you could say, deploy 4 vms instead so add 16 more servers. These vm will be allocated rapdily instead of having to wait x amount of time for each new server to come up.
After x amount of time, if they are not used, the extra vm will shut down and it will try to only fill the standby count you asked for using your vm config.
If there are confusions still, i will answer them to best of my ability.
unique pivot
Wow, with an example it's REALLY easy to understand!! Thanks~
I got stuck since I kept thinking that the "target standby" is just a fixed value and it won't scale, and I need to use Dynamic Standby to scale it. But the conclusion is that, the original version can scale by itself, and Dynamic Standby is just a "scale rule modifier". (Right?)
And it seems "Servers per VM: 25, Target standby: 25" is really stupid since it will directly start another VM.
I guess target standby should be about half? e.g. Target standby: 15, and adjust with Dynamic Standby