Just looking to get some clarification on t2 server's performance over time. We currently are using the legacy PlayFab servers hosted through AWS, on the default t2.medium servers.
- Our game has around 45 players per game session.
- Each game session lasts less than an hour (at 45 minutes the server starts to enter a "world closing state" - forcing all players to escape and disconnect, or die at 50 minutes. Game server instance force closes at 55 minutes). Our games could be fully persistent, but we would like to add a close server state to ensure cleanup happens, logs get captured, etc...
- Game server uses between 2gb and 2.7gb of RAM Testing on t2.medium ourselves, a server instances uses between 50% and 60% CPU Utilization
With these notes, we can only reliably host 1 server instance per t2.medium. With the larger t2.2xlarge we can reliably host 5 server instances.
Obviously hosting a single server instance on a VM isn't ideal - per the notice when setting the "Min free slots" and "Max instances per host" both to 1. If this wasn't a huge issue, we are still worried about hitting the 75000 server hour limit, and the 100 VM instance limit. Can these limits be raised further, through a specific plan type or otherwise?
Switching to the larger t2.2xlarge would help with some of these issues, but my concern then is without being able to reliably close servers after 55 minutes, our t2 credits would run out, and performance would be degraded.
In short I guess my questions are:
- Can the 75000 server instance hours / 100 server instances be increased? Through a custom plan or otherwise?
- Do PlayFab t2 servers run out of credits like other t2 servers? Is there a way to switch to unlimited mode? What happens if no credits remain?
- Is there a way to use both externally hosted, and playfab hosted servers?
Thanks in advance for any insights! Much appreciated.
