Amazon GameLift Servers pricing
Amazon GameLift Servers Instances
Amazon GameLift Servers instance costs are the costs associated with the compute resources you need to host your game, specifically the number of concurrent users (CCU) your game servers need to support. Amazon GameLift Servers supports 68 instance families across 26 regions and 9 Local Zones. Amazon GameLift Servers supports game servers that run on Windows Server 2016, and most recently, Amazon Linux 2023. Windows operating systems include an additional licensing cost, which can increase your instance costs by almost 2x over comparable Linux servers. This cost saving makes the Linux server operating system a good choice for most Amazon GameLift Servers customers. Amazon GameLift Servers offers built in options for reducing your overall game server hosting costs. These include Linux instance types, Amazon GameLift Servers auto-scaling, Spot instances, Graviton instance types, and container packing. By using a combination of these cost-saving options, your game server hosting can be reduced down to $1 per user per month. Here is how each of these can help reduce overall costs.
Typical multiplayer games use 50% of their peak server capacity per region on average. This means that half of the time, your game servers have no active players. Amazon GameLift Servers' intelligent autoscaling algorithm can rapidly scale up and down to save costs while preventing players from waiting by quickly responding to rising player demand. Autoscaling can automatically turn off game servers when they are not needed and adds new game servers when more players arrive.
Amazon GameLift Servers offers both On-Demand and Spot instances. Spot instances offer you the same high-performance of on-demand game servers at savings of between 50% and 85% compared to On-Demand prices. Amazon GameLift Servers uses a proprietary algorithm to place new sessions on game servers to optimize for low player latencies, low instance prices, and minimize Spot interruption rates to help with cost savings while maintaining the player experience. This means that with Amazon GameLift Servers, Spot instances can work well for gameplay sessions of 30 minutes or less or for tutorials modes where a possible interruption is less impactful to the overall player experience. Spot Instance prices adjust gradually based on long-term trends in Spot instance capacity.
Amazon GameLift Servers supports AWS Graviton instances, which are ARM-based processors designed to deliver high performance at a more cost-effective price point than x86-based instances. AWS Graviton instances work well for workloads that involve network processing and data compression, which means they are a great choice for multiplayer game servers. They are available in instance types ranging from general purpose to compute-optimized to memory-optimized. Graviton instances are ideal for all Linux-based workloads written in popular programming languages, such C++, C#, and C, and If you are using the Unreal Engine you can easily build to ARM to take advantage of Graviton instances and cost savings!
Amazon GameLift Servers Containers provide a solution for configuring resource limits per game server, enabling multiple game servers to run on a single Amazon GameLift Servers instance without competing for resources. By understanding the resource requirements of your game server container and the available resources of different instance types, you can determine the optimal number of containers you can pack on each instance.
Network bandwidth is typically one of the largest and least predictable costs for multiplayer game hosting. By eliminating bandwidth charges entirely, Amazon GameLIft Servers you significantly reduces your total infrastructure costs and only charges by GameLift instance hours. No enrollment, pricing agreement, or configuration change is needed. Existing customers on eligible fleets receive the benefit automatically.
Pricing Examples
The following are examples games and calculator-produced Amazon GameLift Servers pricing for each. You can also use the pricing calculator to create a custom pricing example based on your exact requirements.
|
Calculator Input and costs
|
5v5 shooter
|
1v1 Fighter
|
1v1 Card Game
|
100 Player Battle Royale
|
|---|---|---|---|---|
|
Game Engine
|
Custom C++ |
Unreal 4 |
Unity |
Unreal 5.X |
|
Players per game session
|
10 |
2 |
2 |
30 |
|
Bandwidth per player (KB/s)
|
20 |
5 |
2 |
20 |
|
Game sessions or containers per instance
|
8 |
24 |
24 |
4 |
|
Session length (minutes)
|
20 |
5 |
5 |
20 |
|
Operating System
|
Linux |
Linux |
Linux |
Linux |
|
% On Demand
|
100 |
100 |
50 |
100 |
|
% Spot
|
0 |
0 |
50 |
0 |
|
Peak CCU in Month
|
50,000 |
10,000 |
10,000 |
100,000 |
|
Average CCU vs. peak (%)
|
30 |
30 |
30 |
30 |
|
Instance type
|
c6a.4xlarge (8 core, 16 vCPU, 32 GB) |
c6a.4xlarge (8 core, 16 vCPU, 32 GB) |
c6a.large (1 core, 2 vCPU, 4 GB) |
c6i.4xlarge (8 core, 16 vCPU, 32 GB) |
|
Region
|
Us-east (Ohio) |
Us-east (Ohio) |
Us-east (Ohio) |
Us-east (Ohio) |
|
Compute costs per month
|
$111,567 |
$37,440 |
$3,237 |
$165,217 |
|
Bandwidth costs per month
|
$0 |
$0 |
$0 |
$0 |
|
Total cost per month
|
$111,567* |
$37,188* |
$3,237* |
$165,217* |
|
Savings with Graviton
|
|
$35,185
|
|
|
|
Savings with Spot
|
$88,124 (21% savings) |
|
|
$104,671.05 monthly
|
*for an explanation of the total cost, see the expanded examples below
Note: we offer pricing discounts for large scale games, please contact us if your game costs are over $80K per month or just need help estimating your costs.
Game type: 5v5 competitive shooter
An online multiplayer 5v5 competitive shooter game requires a robust cloud infrastructure with specifications like 10 players per session, 20 KB/s bandwidth per player, a 64 tick rate game server running on a custom C++ engine on Linux OS, and a peak of 50,000 concurrent users. For a competitive shooter, results matter since the game cannot tolerate interruptions. We’ll use on-demand instances only and this game requires higher bandwidth per player.Pricing calculator inputs
|
Input
|
Value
|
Formula
|
|---|---|---|
|
Operating System
|
Linux |
|
|
Game engine
|
Custom C++ |
|
|
Peak CCU in Month
|
50,000 |
|
|
Average CCU vs. peak (%)
|
30 |
|
|
Players per game session
|
10 |
|
|
game sessions/containers per instance
|
8 |
|
|
Instance type
|
c6a.4xlarge (8 core, 16 vCPU, 32 GB) @ $0.741/hour |
|
|
% On demand vs % Spot
|
100/0 |
|
|
Bandwidth per player (KB/s)
|
20 |
|
|
Session length (minutes)
|
20 |
|
|
Instance buffer
|
10% |
|
|
Region
|
Us-east (Ohio) |
|
|
Compute costs per month
|
$111,566.81 |
Instance Cost Calculation
|
|
Bandwidth costs per month
|
$0 |
Data Transfer Out (DTO) Cost Calculation
|
|
Total cost/month
|
$111,566.81 |
Instance Cost (monthly) $111,566.81 + Data Transfer Out (DTO) Cost (monthly): $0 = Total Cost (monthly): $111,566.81 |
|
Spot instances + bandwidth total cost/month
|
$88,124 (21% savings) |
For the same competitive 5v5 online shooter game, strategically utilizing 30% AWS Spot instances for lower-priority shorter session game modes or firing ranges reduces the estimated monthly infrastructure cost by over $23,443, a significant savings without compromising the core multiplayer experience. |
Note: we offer pricing discounts for large scale games, please contact us if your game costs are over $80K per month or just need help estimating your costs.
Game type: 1v1 Fighting
An online multiplayer 1v1 fighting game, like the classic Street Fighter series, requires a lightweight cloud infrastructure with 2 players per session, 3 KB/s bandwidth per player, a 60 tick rate Unreal Engine 4 game server running on Linux, and utilizing cost-effective c5a.4xlarge instances to support short, uninterrupted combat sessions.
|
Input
|
Value
|
Formula
|
|---|---|---|
|
Operating System
|
Linux |
|
|
Game Engine
|
Unreal 4 |
|
|
Peak CCU in Month
|
10,000 |
|
|
Average CCU vs. peak (%)
|
30 |
|
|
Players per game session
|
2 |
|
|
game sessions/containers per instance
|
24 |
|
|
Instance type
|
c6a.4xlarge (8 core, 16 vCPU, 32 GB) @ $0.741/hour |
|
|
% On demand vs % Spot
|
100/0 |
|
|
Bandwidth per player (KB/s)
|
3 |
|
|
Session length (minutes)
|
5 |
|
|
Instance buffer
|
10% |
|
|
Region
|
Us-east (Ohio) |
|
|
Compute costs per month
|
$37,188.94 |
Instance Cost Calculation
|
|
Bandwidth costs per month
|
$0 |
Data Transfer Out (DTO) Cost Calculation
|
|
Total cost/month
|
$37,188.94 |
Instance Cost (monthly) = $37,439.88 Data Transfer Out (DTO) Cost (monthly) = $0 Total Cost (monthly) = $37,188.94 |
|
Graviton instance + Bandwidth total cost/month
|
$35,185.37
|
Using c6g.4xlarge (8 core, 16 vCPU, 32 GB) @ $.66/hour vs c5a.4xlarge (8 core, 16 vCPU, 32 GB) @ $0.746/hour |
Note: we offer pricing discounts for large scale games, please contact us if your game costs are over $80K per month or just need help estimating your costs.
Game type: 1v1 Card game
An online multiplayer 1v1 card game that only needs a lightweight infrastructure with 2 players per session, 2 KB/s bandwidth per player, a 10 tick rate Unity game server on Linux leveraging cost-effective c5a.large instances to facilitate short, uninterrupted sessions.
|
Input
|
Value
|
Formula
|
|---|---|---|
|
Operating System
|
Linux |
|
|
Game engine
|
Unity |
|
|
Peak CCU in Month
|
10,000 |
|
|
Average CCU vs. peak (%)
|
30 |
|
|
Players per game session
|
2 |
|
|
game sessions/containers per instance
|
24 |
|
|
Instance type
|
c6a.large (1 core, 2 vCPU, 4 GB) @ $.099 OD /$.03 Spot |
|
|
% On demand vs % Spot
|
50/50 |
|
|
Bandwidth per player (KB/s)
|
2 |
|
|
Session length (minutes)
|
5 |
|
|
Instance buffer
|
10% |
|
|
Region
|
Us-east (Ohio) |
|
|
Compute costs per month
|
$3,237.09 |
Instance Cost Calculation
|
|
Bandwidth costs per month
|
$0 |
Data Transfer Out (DTO) Cost Calculation
|
|
Total cost/month
|
$3,237.09 |
Total Cost Instance Cost (monthly): $3,237.09 +
|
Note: we offer pricing discounts for large scale games, please contact us if your game costs are over $80K per month or just need help estimating your costs.
Game type: 100 Player Battle Royale
Battle royale games see significant player counts decrease after initial "hot drop" phases, so and on-demand solution optimizes for an average of 30 concurrent players per session rather than the full 100 player capacity, utilizing the high-performance Unreal Engine 5.X on Linux instances to deliver seamless online experiences.
|
Input
|
Value
|
Formula
|
|---|---|---|
|
Operating System
|
Linux |
|
|
Game Engine
|
Unreal 5.x |
|
|
Peak CCU in Month
|
100,000 |
|
|
Average CCU vs. peak (%)
|
30 |
|
|
Players per game session
|
30 |
|
|
game sessions/containers per instance
|
4 |
|
|
Instance type
|
c6i.4xlarge (8 core, 16 vCPU, 32 GB) @ $.823 |
|
|
% On demand
|
100 |
|
|
Bandwidth per player (KB/s)
|
20 |
|
|
Session length (minutes)
|
20 |
|
|
Instance buffer
|
10% |
|
|
Region
|
Us-east (Ohio) |
|
|
Compute costs per month
|
$165,217.25 |
Instance Cost Calculation
|
|
Bandwidth costs per month
|
$0 |
Data Transfer Out (DTO) Cost Calculation
|
|
Total cost/month
|
$165,217.25 |
Instance Cost (monthly): $165,217.25 + Data Transfer Out (DTO) Cost (monthly): $0 = Total Cost (monthly): $165,217.25 |
|
Graviton instance + Bandwidth total cost/month
|
$215,226 monthly
|
Using Graviton c6g.4xlarge (8 core, 16 vCPU, 32 GB) instance @ .66/hour vs c6i.4xlarge (8 core, 16 vCPU, 32 GB) @ $.904 |
|
Spot + Graviton instance + Bandwidth total cost/month
|
$104,671.05 monthly
|
Using Graviton c6g.4xlarge (8 core, 16 vCPU, 32 GB) instance @ .66/hour for 70% of workload and Graviton Spot @ $.2058 for the remaining 30% |
Note: we offer pricing discounts for large scale games, please contact us if your game costs are over $80K per month or just need help estimating your costs.