Uken Games Slashes Observability Costs 87% with ClickHouse

Open Source Observability Wins Big

Uken Games' transition to a ClickHouse-based observability stack offers compelling evidence for the power of open-source solutions in reducing costs and mitigating vendor lock-in. The 87% cost reduction is a significant achievement, particularly for a company serving millions of players, and highlights the potential for substantial savings by moving away from proprietary, often complex, pricing models like Datadog's. The adoption of OpenTelemetry as a vendor-agnostic data plane is a forward-thinking strategy, ensuring flexibility and future adaptability. The pragmatic approach to data retention and sampling, focusing on actionable insights rather than exhaustive data collection, is a crucial lesson for any organization looking to optimize their observability pipelines. The success hinges on the robustness and performance of ClickHouse, which appears to have delivered exceptionally well, handling Uken's entire trace volume on a single node.

However, the article, while strong, could benefit from a deeper dive into the specific technical challenges encountered during the migration and the strategies employed to overcome them. While Alexei Zenin mentions "re-engineering and building" other components, the specifics of how the OpenTelemetry collectors were tuned, how SigNoz was integrated, and how Grafana queries were optimized for ClickHouse performance would add significant value for engineers considering a similar path. Furthermore, while the article emphasizes cost savings, a brief mention of the total infrastructure costs post-migration, even if estimated, would provide a more complete financial picture. The decision to keep logs in CloudWatch, citing cost and engineering effort, is understandable but could be explored further regarding potential future cost savings if logs were also migrated to an open-source solution.

Ultimately, this case study is a powerful endorsement for the open-source observability ecosystem, particularly for companies experiencing sticker shock from commercial vendors. It demonstrates that high performance, scalability, and significant cost savings are achievable with the right combination of open-source tools. The focus on pragmatic data management and the ability to achieve feature parity with established commercial solutions makes this a highly relevant and influential example for the industry. The article's strength lies in its clear articulation of the problem, the elegant solution, and the quantifiable results, making it a valuable read for DevOps engineers, SREs, and CTOs alike.

Key Points

• Uken Games replaced Datadog with an open-source, ClickHouse-based observability stack, achieving an 87% reduction in observability costs.
• The new stack utilizes OpenTelemetry as a vendor-agnostic data plane, collecting metrics, traces, and logs.
• ClickHouse serves as the trace store, handling all of Uken's trace data efficiently on a single node.
• SigNoz is used for APM functionality on top of ClickHouse for traces, providing feature parity with Datadog.
• Prometheus (AWS-managed) is used for metrics, and Grafana for visualization and alerting.
• The migration involved a pragmatic approach to data retention and sampling, focusing on actionable insights and dropping unnecessary data.
• The company achieved full control over its infrastructure and avoided vendor lock-in.

---

📖 Source: How Uken Games reduces observability costs by 87% with ClickHouse