Game Metrics Cloud Gem

You can use the Game Metrics cloud gem to collect event data on player behaviors, which you can then analyze or trigger event actions. This gem uses a service API with the Cloud Gem Framework for event submission, Amazon S3 for data storage, Amazon Athena as the query engine, and Amazon QuickSight as the data visualizer.

You can enable this gem to do the following:

• View realtime data about users playing the game.
• Run queries and generate reports on player behavior over time.
• Visualize player activity in the game world.
• Record player events, even if players are offline.

Note
The Game Metrics cloud gem is only supported in regions where FIFO is supported. For more information, see Amazon SQS FIFO (First-In-First-Out) Queues in the Amazon Simple Queue Service Developer Guide.

Example Overview
The following illustrates how AWS services process game events.

When events are generated from your game, the following occurs:

1. Events submitted by REST requests are first processed by the Amazon API Gateway Lambda handler called the FIFO_producer (producer). This producer breaks the request into smaller chunks in order to upload a maximum of 256KB per chunk, which is the maximum payload size an SQS message can support. Because the size of a single event, including attributes, can’t exceed 256KB, the producer returns an error if an event exceeds this limit.

2. A Lambda process called the FIFO_consumer (consumer) processes the SQS messages.

3. The consumer is triggered every five minutes and self-replicates when one of the following conditions are met:

   • SQS queue growth exceeds 5%, which is checked every 15 seconds.
   • SQS queue count exceeds 3000 messages.

   The consumer processes as many SQS messages as possible within its five minute lifespan, aggregates the events of the same type into a single parquet file, and then saves the file to a specific S3 key.

4. Another Lambda function called the amoeba_generator is triggered every 20 minutes and aggregates all files in a single key path into a single file. This process maximizes IO (input/output) performance when loading the files by the database. AWS Glue crawlers crawl the data bucket and identify a common schema; this schema defines the tables for Athena to use.

5. The S3 files are made available to Athena for querying through AWS Glue. AWS Glue updates Athena’s database schema every hour. The S3 event data is partitioned (or indexed) by the hour. This means the lowest level of granularity when using Athena partitions is by the hour. Note
   Sending queries based on partitions can greatly reduce costs and improve query performance. For more information, see Partitioning Data in the Amazon Athena User Guide.

The Game Metrics cloud gem includes the following default Athena partitions: