GAMESUS01-BP01 Use storage technologies that fit the patterns adapted to user content, subscriber information, and in-game purchases

You should classify your data by type, retention need and frequency of access. This enables you to select the most optimized storage solution for the myriad data types of your game or backend services produce. Fast changing data should be stored in key-value or in-memory database services. Transactional data should be store in relational database services. Large files, game assets, or user-generated content should be stored in object storage services.

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

Implementation guidance

Games produce and consume a large variety of data types that require storage solutions optimized for frequency of access, latency, and cost. Data stored should be classified using tags to differentiate data that can be removed or needs to be stored long-term.

The following services work well for a variety of Games use-cases:

Amazon Aurora (compatible with MySQL and PostgreSQL) offers high availability, low-latency, and automatic scaling, making it an excellent choice for handling large amounts of transactional data, such as player account management and authentication, in-game economies, leaderboards and player rankings, game state persistence, event and campaign management, and multi-Region and high-availability deployment.

Amazon DynamoDB is a fully managed NoSQL database known for its low-latency, high throughput, and seamless scalability, which makes it ideal for handling real-time player data, session management, inventory, in-game economy, real-time multiplayer game state, matchmaking, event logging, and scaling for global audience.

Amazon DocumentDB (compatible with MongoDB) provides a scalable, low-latency document-oriented database service, perfect for storing flexible, semi-structured data, such as inventory system, player profiles and customization's, game worlds and procedurally generated content, social and player interactions, analytics and behavior tracking, and in-game metadata and configurations.

Amazon ElastiCache supports in-memory caching with Redis or Memcached, offering rapid data access and reduced response times, which is critical for real-time multiplayer games where speed and performance are essential for a smooth user experience. ElastiCache is utilized in gaming for real-time leaderboards, session management, caching game metadata, in-game chat and messaging, matchmaking, real-time analytics and telemetry, and scaling for high-traffic events.

Amazon Simple Storage Service (S3) can be used to store objects like game assets, videos, pictures, text log files and more. S3 is an object storage service offering industry-leading scalability, data availability, security, and performance.

If it offers multiple storage classes that support frequent and in-frequent data access, and cost-effective archive storage. For data that is frequently accessed throughout development, studios should store objects in S3 Standard for low latency and high throughput performance. For data that frequently goes from hot to cold or vice versa, studios should investigate S3 Intelligent-Tiering. Intelligent-Tiering monitors the access patterns of your data and automatically moves data to the most cost-effective access tier.

For studios that need high throughput, low latency and are ok with it living in a single Availability Zone use S3 Express One Zone. This replicates data to a single AZ and can improve data access speeds compared to S3 standard. For deep archive needs of historical data Amazon also offers Amazon Glacier. The Amazon Glacier storage classes are purpose-built for data archiving, providing you with high performance, retrieval flexibility, and low cost archive storage in the cloud.

Amazon Elastic Block Store can be used to store game servers' binaries, executable files and configurations your game servers or asset repositories need to function. You should snapshot and delete unused volumes that are not attached to an EC2 instance. This alleviates you from storage charges incurred while lowering the usage of unneeded services and hardware.

Implementation steps