Now that you have hands on experience building a batch and real time analytics pipeline and integrating it into a Unity game, lets talk about how you can extend that pipeline. You can build upon your analytics pipeline to create event-driven applications in response to your actionable insights, use machine learning, and even do live ops monitoring of your game.
Live-ops is the ability to create an event-driven architecture that helps you monitor and manage your infrastructure to support three main areas: community building, measuring and responding to player behavior, and adding new content and features to your game.
For example, with live ops you can better foster a community by making sure your infrastructure is well equipped to handle live events or social campaigns, using real-time analytics to monitor these events, and creating an event-driven architecture that better allows you to respond to these events in real-time.
Another important component of live ops is the ability to push new content to your game to increase player engagement. This is is where a Continuous Integration / Continuous Deployment (CICD) pipeline is helpful so that you can take what you have built, distribute it to your players quickly, and monitor the entire process along the way with analytics.
To learn more about live ops analytics, click here.
You can create event-driven applications to take action based on the insights you are collecting from your analytics pipeline. You can do event-based fan out and notification with Amazon SNS (Simple Notification Service), which is a fully managed pub/sub messaging service that enables you to decouple microservices, distributed systems, and serverless applications. For example, you can configure your pipeline to push messages to SNS to publish alerts and anomalies to a topic which is used to notify users and other applications.
You can also use Amazon SQS (Simple Queue Service) to help create event-driven architectures as well, which helps decouple services and facilitate the communcication between downstream applications.
Another option is using CloudWatch Events or Amazon EventBridge. You can set up rules that respond to operational changes in your environment and take corrective action as necessary, by sending messages to respond to the environment, activating functions, making changes, and capturing state information. For example, if your CPU utilization for an EC2 instance surpasses a certain threshold, you can create a CloudWatch event that triggers your compute to scale up.
To learn more about creating event-driven architectures, click here.
Another way you can extend your analytics capabilities is through the use of machine learning. From detecting fraud and predicting player behavior, to automating playtesting, machine learning can make your game development processes faster and smarter. It can help you create more engaging games by making your games feel more alive, changing the way that players interact with your games. It can also help with anomaly detection and fraud detection to detect and prevent cheating. You can use machine learning for revenue generation to better monetize your game by incorporating personalized content recommendations. It is also helpful for quality assurance and level difficulty analysis, where you can create bots that can play through your level design to see if it needs to be improved.
To learn more about machine learning for games, click here.