View a markdown version of this page

GAMEPERF01-BP02 Consider operational overhead for scaling game servers - Games Industry Lens
Implementation guidance

GAMEPERF01-BP02 Consider operational overhead for scaling game servers

Consider the management and operational overhead associated with each hosting option.

Level of risk exposed if this best practice is not established: High

Implementation guidance

Operational overhead

Self-hosted solutions on EC2 or containers can provide more control but also will require more management. Container orchestrators like ECS or EKS can reduce launch times for containerized servers while also increasing networking complexity and maintenance orchestration overhead.

As an example, EKS managed node groups can automate provisioning and lifecycle management of your game servers but do not respect pod disruption budgets when terminating a node, if your game requires longer than the 15 minute termination period to safely complete games, you may need to create lifecycle hooks or consider self-managed nodes with custom controllers to block game interruption.

Managed services like Amazon Game Lift may handle most of the operational overhead but reduce the amount of visibility and control over special requirements for low level networking and security configuration. Choosing a game server solution is a trade-off between the level of customization, control and responsibility you will have for tuning game server performance and scaling behavior.

Implementation steps

  • Assess operational overhead for hosting options, balancing control and management effort between self-hosted solutions like EC2, ECS, or EKS and managed services like Amazon Game Lift.

  • Use EKS managed node groups for automation but implement lifecycle hooks or custom controllers if your game servers require longer termination periods than the default.

  • Weigh the trade-offs between customization, visibility, and operational responsibility when selecting a game server solution.