How PerfectScale helps a DevOps team of one scale at Riftweaver
Meet Riftweaver
Riftweaver is an innovative gaming studio that brings tabletop role-playing (RPG) gaming to users on mobile devices, and it has made big strides recently. From launching its flagship game to introducing gameplay on platforms like Steam and iPads, Riftweaver’s game has over 80,000 downloads. With a small team of just 15 employees, Riftweaver’s backend infrastructure is managed by a single DevOps engineer, Rumby Osei,who ensures the game’s infrastructure performance, efficiency, and scalability – all while delivering a seamless experience for its growing player base.
The Challenge
“When I joined Riftweaver, the game wasn’t scaling or handling production-ready load. My role was to get us there.” This was the initial remit of the DevOps function at Riftweaver – and Osei was eager to address it.
As the Riftweaver gaming studio prepared to move from closed beta to early access and public launch, Osei knew that unpredictable player load would immediately pressure the product to scale without sacrificing reliability.
This cloud-native modern apps gaming provider runs. With a production cluster and multiple non-production environments running on AWS EKS, Osei relied on Cluster AutoScaler (CAS) for node scaling and Horizontal Pod Autoscaler (HPA) for pod scaling. However, when some services began to show CPU throttling, she turned to Kubernetes Vertical Pod Autoscaler (VPA) but hit roadblocks due to missing historical context and inconsistent recommendations.
“VPA didn’t provide historical data or let me generate consistent results. I couldn’t go back and see what had changed over time. I even had to write a script to pull down all the VPAs for each deployment just to compare data.”
This created several key challenges for Riftweaver, including:
- CPU throttling
Frequent throttling during load tests and live gameplay, leading to game lag and poor user experiences. - Time-consuming troubleshooting
Performance analysis was time-consuming and manual, often producing conflicting signals. - Limited optimization
Existing tools like VPA lacked historical insight and prescriptive guidance, making optimization slow and risky.
The Solution
Rumby Osei sought a solution to safely right-size with data-driven guardrails, reduce performance bottlenecks, and provide actionable, workload-level insights into resource usage. After a discussion with PerfectScale by DoiT’s team, Rumby saw its potential to bridge the gap between load-testing predictions and real-world user behavior to improve SLO-aligned resiliency.
PerfectScale by DoiT’s intuitive interface, robust analytics, and evidence-backed recommendations made it an attractive tool for Riftweaver’s small but dynamic team. Features like environmental impact (CO₂) tracking and HPA visibility aligned with the studio’s forward-thinking approach to infrastructure management.
With PerfectScale, Riftweaver can now address its needs for performance, efficiency, and scalability in several key ways:
- Reduce CPU throttling: PerfectScale identified high-risk workloads and provided tailored, low-risk recommendations for CPU adjustments.
- Simplify resource allocation: Insights into underutilized nodes and workloads enabled Osei to optimize pod placement and node selection, reducing waste and improving efficiency at scale.
- Improve resiliency By prioritizing changes based on PerfectScale’s risk profiles, Riftweaver reduced dropped/duplicate API calls, improving player experience.
These outcomes make PerfectScale an integral part of Osei’s infrastructure strategy. “PerfectScale’s UI and risk profiles gave me confidence in the recommendations. It also introduced features I hadn’t seen before—like monitoring environmental impact with CO2 tracking.”
The Results
Improvements like these are foundational to helping Riftweaver scale efficiently as it meets growing demand.
Key outcomes include
- 59% reduction in CPU throttling
Six critical APIs showed significant improvements, ensuring smoother gameplay for users. - Improved user experience
Players experienced fewer interruptions, even during peak usage, reinforcing Riftweaver’s commitment to quality. - Streamlined operations for a one-person DevOps team
Osei gained the ability to proactively manage scaling and resiliency with minimal manual effort. - Cost-effective scaling
s part of PerfectScale’s Community Package, Riftweaver accessed these capabilities at no cost, optimizing infrastructure without budget strain.
“PerfectScale has been a game-changer. The insights have not only solved immediate issues but also prepared us for future growth as we expand to more platforms and users.”
By leveraging PerfectScale by DoiT, Riftweaver optimized its Kubernetes environment, ensuring scalability, reliability, and a consistently superior player experience. As the studio continues to grow, PerfectScale remains an important partner in its journey to deliver exciting, lag-free gameplay to players worldwide.
