CIG is a highly innovative business, not only in its ambitious approach to game development, but also in its use of a series of record-breaking crowdfunding campaigns to finance the initial creation of Star Citizen and Squadron 42. When Star Citizen entered Playable Alpha at the end of 2015, CIG continued to leverage its direct-to-consumer model to grow its player base and increase revenues on its way to its current Early Alpha Access stage.
The direct-to-consumer model gives CIG an unfiltered relationship with its customers that most traditional game studios do not have. Nearly all the project’s early backers are players themselves and are contributing to the development of Star Citizen and Squadron 42 by play-testing pre-release builds of the games.
CIG monitors and analyzes these players’ in-game behavior to understand which aspects of the game are driving engagement, and where improvements are needed—helping to test design decisions and fine-tune gameplay.
Miles Lee, Director, Build Operations at Cloud Imperium, explains: “It’s a case where actions speak louder than words. Detailed analysis of player behavior is important to everyone at Cloud Imperium, from our game designers and engineers to the marketing team.”
Building an analytics platform for a massively multiplayer online game is a significant technical challenge because player activity generates vast numbers of data points every second.
Initially, CIG built a pipeline using a Function as a Service (FaaS) platform and Python scripts to handle the data flows, but this solution was difficult to maintain and suffered performance issues as volumes increased. Moreover, with business users requesting new metrics all the time, it was also increasingly difficult for the data team to keep up with demand.
Equally, the FaaS platform was becoming an expensive option, especially when new releases of the latest Star Citizen builds led to sudden spikes in the number of players online.
CIG’s Development Operations (DevOps) team had used Scala as its preferred programming language for more than a year and was an enthusiastic proponent of Akka Platform’s reactive architecture for developing reliable, scalable event-driven architectures. For example, CIG DevOps used Akka Platform to develop web apps and frameworks used internally to improve developer experience.
As a result, when CIG decided to rearchitect its analytics platform, Lightbend’s open source Cloudflow project made immediate sense, because it is a streaming platform that supports both Akka and other streaming engines such as Apache Spark and Flink.
As its use of Cloudflow took off, CIG signed up to Akka Data Pipelines with Lightbend. Akka Data Pipelines is the subscription version of Cloudflow, providing additional functionality as well as comprehensive support from Lightbend’s experts.
Miles Lee explains: “The support we get from Lightbend is phenomenal—whenever we need help, we just raise a ticket and get our issues resolved quickly.”
Akka Data Pipelines makes it simpler to develop, deploy, and manage pipelines as an “always on” service.
Miles Lee comments: “Akka Data Pipelines provides great deployment tools that made it easy to stand up the infrastructure for our data pipelines quickly. We only needed one engineer on the project to get up and running, which is important for our relatively small team.”
The new architecture allows Star Citizen game servers to send player behavior data into both AWS Kinesis for real-time processing, and Kinesis Data Firehose, which ingests historical data into an S3 data lake. Downstream, Akka streamlets managed by Akka Data Pipelines handle both real-time processing straight from Kinesis and triggered reprocessing from the S3 data lake, in the event of downstream data loss or when business decisions make it necessary to handle historical data in a new way.
The decision to use Akka Platform and Akka Data Pipelines from Lightbend has not only provided a robust model for building reliable, scalable platforms—it has also helped CIG attract new talent.
“People are excited to work with Scala and Akka at a game company,” says Miles Lee. “Here at CIG, there’s huge scope for using technology in exciting and innovative ways, so it’s easy to hire smart people with the right skills.”
With Akka Data Pipelines, CIG has a powerful platform for extracting insight from player behavior data, helping designers and marketing teams make smarter decisions. Moreover, CIG can now trust its infrastructure to react predictably to scaling challenges.
“We no longer have to worry about scaling issues, which is great!” says Miles Lee. “We’re also seeing a much faster turnaround on new analytics requests. If marketing wants to track a new metric, we can create a new streamlet, replay the last six months of data, and get them an answer quickly.”
CIG is now working on an Akka Platform implementation of its build system, which will allow developers to test builds without recompiling the entire codebase. This will enable faster iterations and help push new features out to players faster than ever before.
Founded in 2012, Cloud Imperium Games (CIG) is the game studio behind Star Citizen, a massively multiplayer online space sim, and Squadron 42, a single-player adventure set in the Star Citizen universe. CIG employs more than 700 people at five offices around the globe.