News for hackers. Stuff that matters.
Re:Postmortem of Service Outage at 3.4M Concurrent Users by Thaxll on Friday February 09, 2018 @01:33PM (#16342664)
Since gameservers are mostly single threaded they probably run 35~~ of them on a single instance ( c4.8xlarges have 36 vcores ), so if you have a good matchmaking algorithm you can fit 3500~~ players per instance, to reach 3.5M you need between 1000 and 2000 instances.

As for your last comment it's wrong, a game like Battlefield is probably more demanding compare to a battle royale type of game, a FPS like Battlefield runs at 60hz and is using a lot of physics everywhere at fast pace, battle royal games don't have that much action going on.

Re:Postmortem of Service Outage at 3.4M Concurrent Users by Flammy on Friday February 09, 2018 @01:49PM (#16342835)
Likely more than 35 games on a single instance, as ideally you wouldn't be starting all games at the same time on the same instance. Staggering the start of games on each instance would be ideal.

For example you could have dozens of late stage games that combined have less players than a single, just started, 100 person game.

Re:Postmortem of Service Outage at 3.4M Concurrent Users by evanosaurus on Friday February 09, 2018 @02:08PM (#16342980)
We run 16 game servers per c4.8xl instance for Battle Royale. We run more than that (on a different instance type) for StW.
Re:Postmortem of Service Outage at 3.4M Concurrent Users by Thaxll on Friday February 09, 2018 @02:00PM (#16342926)
You usually don't run two or more game server on single a thread / core, it kills performance, and with VMs with many physical processor you have other issues like NUMA so you have to be very careful.

I'm pretty sure they have a ratio of one gameserver per thread / core.