Photo credit: venturebeat.com
Amazon Web Services (AWS) has announced that its Amazon GameLift service has achieved a benchmark performance capable of supporting up to 100 million concurrent users (CCU) for a single game.
Through the use of Amazon GameLift, AWS can significantly reduce the server hosting challenges faced by game developers by dynamically adjusting backend resources as needed. The testing revealed that the platform can accommodate an additional 100,000 players every second and initiate over 9,000 new compute instances each minute. This provides developers with a scalability potential that surpasses previous limits.
Amazon GameLift functions as a cloud-based solution that assists developers in integrating game servers for multiplayer experiences, built on the robust infrastructure of AWS.
The company recognizes the immense dedication and resources that online game developers invest in creating new titles and updates. On launch days, the true test of these efforts occurs as players rush to access the game, often without clear expectations of user demand. Nonetheless, developers understand that their backend infrastructure must accommodate this influx with seamless scalability to ensure player satisfaction and interaction remain intact.
Chris Byskal, general manager of AWS Game Services/Game Tech, shared insights, noting, “Although most developers may not disclose CCU figures, informal tracking shows that the leading games today peak around 14 million CCU, with the entire Steam platform reaching nearly 40 million CCU. Our target of 100 million CCU is set as a benchmark to demonstrate how Amazon GameLift can effortlessly support even the largest titles, far exceeding current demands.”
For developers, estimating CCU serves as a foundational step in planning for scaling needs leading up to a game’s release. Additional factors for scaling include managing sudden traffic spikes, handling game session allocations, and considering the geographical distribution of players.
The team behind Amazon GameLift boasts nearly ten years of experience in helping game creators effectively navigate the challenges of large-scale launches while minimizing errors. This expertise has been honed through testing the solution in extreme conditions. Further resources, including a detailed walkthrough for scaling to 10 million and 100 million CCU, are available through demonstrations provided by AWS.
Preparing for Scale
Developers can utilize projected CCU to gauge the necessary virtual machine (VM) capacity needed for optimal gameplay. For instance, to calculate the VM requirements to support a CCU of 10 million, with each VM capable of handling 12 game sessions before performance degradation occurs, the formula would be as follows:
Total VMs = CCU / (players per game session x game sessions per VM)
Total VMs = 10,000,000 / (10 x 12) = 83,333
With availability in 23 AWS regions and nine local zones, Amazon GameLift can efficiently manage the 83,333 VMs required for 10 million CCU. The system can activate virtual machines within minutes, allowing developers to swiftly scale from zero to ten million players. This autoscaling capability is crucial in reducing idle capacity, thereby keeping operational costs in check.
Creating Game Sessions Quickly and Sensibly
AWS effectively allocates players to available game server resources. Each game session must operate on a dedicated game server process, and it’s essential to avoid multiple games sharing the same process. Consider the following calculation for allocation throughput:
Allocation requests/second = CCU / (players per game session x seconds per game session)
Allocation requests/second = 10,000,000 / (10 * 900) = 1,112 (rounded up)
Other critical factors include:
Game sessions: Each session must be linked to a distinct game process.
Location: Servers should be geographically positioned to ensure low latency for players.
Cost: Prioritize the most affordable hardware options when feasible.
Capacity: Use autoscaling to avoid unnecessary expenditures on excess capacity.
Virtual machine usage: Monitor active VMs to prevent accidental termination of ongoing games.
Capacity health: Ensure that allocated infrastructure is operationally sound (e.g., avoiding non-responsive servers).
Number of VM game sessions: Overloading a VM with too many game sessions simultaneously can exceed its computational limits, particularly when resource-intensive tasks are involved.
These considerations can guide the choice of appropriate VMs for specific game sessions. At high volumes, scenarios can arise where numerous game sessions require activation simultaneously within a single AWS region. Attempting to allocate these onto a limited number of idle game server processes may lead to significant connection failures, worsening as more players arrive concurrently.
While these challenges may seem manageable at lower throughput rates, they can escalate quickly, resulting in longer wait times for game initiation or a complete breakdown of availability, which could severely impact the game’s launch.
Scaling with Amazon GameLift
Once developers have incorporated their game server with Amazon GameLift, they need to upload the game server executable along with any required assets to create a Fleet that runs within the AWS infrastructure.
Source
venturebeat.com