Preparing your game for launch with Amazon GameLift hosting

[Critical] Verify that you've completed all the development roadmap steps for your hosting solution, and that you have all the required components in place, including and integrated game server, a backend service for game clients, hosting fleets, and a game session placement method (such as a queue).

[Critical] Create Amazon Identity and Access Management (IAM) roles that allow your game server to access other Amazon resources while running.

[Critical] Design and implement failover to other hosting resources as needed.

Plan the rollout of fleets to your target locations, considering your game's queue and fleet structure.

Automate your deployment using infrastructure as code (IaC) with Amazon CloudFormation and the Amazon Cloud Development Kit (Amazon CDK).

Collect logs and analytics using Amazon CloudWatch and Amazon Simple Storage Service (Amazon S3).

[Critical] Request increases for Amazon GameLift service quotas and other Amazon Web Services service quotas so that your live environment can scale up to production needs.

[Critical] Verify that the open ports on live fleets match the range of ports that your servers could use.

[Critical] Close RDP port 3389 and SSH port 22.

Develop a plan for the DevOps management of your game. If you're using Amazon CloudWatch Logs or Amazon CloudWatch custom metrics, define alarms for severe or critical problems on the server fleet. Simulate failures and test the runbooks.

Verify that the compute resources you're using can support the number of server processes that you want to run concurrently on each compute.

Tune your scaling policy to be more conservative at first and provide more idle capacity than you think you need. You can optimize for cost later. Consider the use of target-based scaling policy with 20 percent idle capacity.

For FlexMatch, use latency rules to match players who are geographically near each other. Test how this behaves under load with synthetic latency data from your load test client.

Load test your player authentication and game session infrastructure to see if it scales effectively to meet demand.

Verify that a server left running for several days can still accept connections.

Raise your Amazon Web Services Support plan level to Business or Enterprise so that Amazon Web Services can respond to you during problems or outages.

[Critical] Set the fleet protection policy to full protection on all live fleets so that scaling down doesn't stop active game sessions.

[Critical] Set fleet maximum sizes high enough to accommodate peak anticipated demand, at minimum. We recommend that you double your maximum size for unanticipated demand.

Encourage your entire development team to participate in the launch event and monitor your game launch in a launch room.

Monitor player latency and player experience.

Tune scaling policy to minimize idle capacity based on player usage.

Modify FlexMatch rules or add hosting locations based on player latency data and revised requirements.

Optimize the runtime configuration to run as many games sessions as possible on each computing resource. Mazimizing performance efficiency in this way can directly affect your fleet costs, because you might be able to run more server processes with the same compute resources.

Use your analytics data to drive continued development, improve player experience and game longevity, and optimize monetization.