Collecting client events (option with AWS SDKs):

Serverless real-time analytics for games

A game, for example, a mobile or PC game, recieves
1 temporary AWS credentials from the Amazon Cognito
identity pool to access Amazon Kinesis Data Streams.

Build serverless, real-time analytics pipelines for cross-platform games 2

The game uses an AWS SDK to submit events in JSON
directly to Kinesis Data Streams. The SDK used depends
This serverless architecture example collects events from games, analyzes them in real-time, and stores them for analysis. It can be on the game engine. If the game engine is Unity, then
you can use the AWS SDK for .NET. If the game engine is
leveraged to provide real-time flash offers, monitor user acquisition campaigns, detect abusive players, find deficiencies and Unreal Engine, then you can use the AWS SDK for C++.
anomalies during A/B testing, and build online dashboards for business metrics such as retention or operational metrics like crashes. Collecting client events (option without AWS SDK):
The game submits events via the REST API to the
3 Amazon API Gateway, which has a native integration
with Kinesis. This option comes at an added cost of
having an API Gateway, but creates an additional layer of
1 separation from the player and the Kinesis Data Stream
via a REST API and does not require Amazon Cognito.
5 6 7
Submitting server events:
Amazon Cognito Multiplayer game servers, backend servers, and other
2 Amazon Simple 4 services can submit events directly to Kinesis using the
Amazon Kinesis AWS Lambda Amazon AWS SDK, or through the API Gateway. When possible,
Game client Data Analytics CloudWatch Notification Service submit events from an authoritative server.

with AWS SDKs (Amazon SNS) Processing and analyzing real-time events:
Real-time analytics Kinesis Data Analytics consumes data from the Kinesis
5 Data Stream instance and allows real-time SQL queries to
run on the stream to analyze, filter, and process data.
The data is processed by a Lambda function, which
3 8 9 6 sends custom metrics to Amazon CloudWatch.

10 The custom CloudWatch metrics are visualized in a real-

Game client Amazon Kinesis Amazon Simple Storage Amazon Amazon 7 time dashboard. You can create an operational
Amazon API Gateway Amazon Kinesis dashboard that shows the health of your infrastructure,
with REST API Data Firehose Service (Amazon S3) Athena QuickSight and a game events dashboard that shows real-time game
Data Streams KPIs such as concurrent users. You can create alerts and
notifications through CloudWatch and SNS.
Storing events for batch analysis:
Amazon Kinesis Data Firehose consumes the data stream
8 and pre-processes the data for storage using a built-in
4 Lambda integration. For example, you can pre-process the
data at this step by transforming it to Parquet, which is a
Game servers compressed format optimized for performance and lower
storage costs when doing analytics.
and backend AWS Lambda Processed data is batched and stored in Amazon S3 to
services Batch and ad hoc analytics 9 create a game analytics data lake. Leverage S3
Intelligent-Tiering, lifecycle policies, and different
storage tiers for cost savings for historical data.
Visualizing batch data:
Reviewed for technical accuracy August 9th, 2021 Query data on an ad hoc basis using Amazon Athena.
© 2021, Amazon Web Services, Inc. or its affiliates. All rights reserved. AWS Reference Architecture 10 Visualize the data to get business insights using Amazon
QuickSight.