Cloudflare has disclosed details of an internal unified data platform named Town Lake, alongside an AI analytics agent called Skipper that provides unified access to operational, billing, security, and business data. The disclosure was made in a blog post by Cloudflare engineer Leela Kumili.
The platform has processed approximately 91,000 billing queries, with billing workloads accounting for 53 percent of all queries, representing the majority of usage. This indicates that billing data is the primary driver of analytics activity on the platform.
Architecture and Core Components
Town Lake is built on a lakehouse architecture that combines the Trino distributed SQL query engine with the Apache Iceberg table format, Cloudflare’s own R2 object storage, and DataHub for metadata management. This stack allows the platform to support governed cross-system analytics across disparate data sources.
The lakehouse approach merges the flexibility of a data lake with the reliability and performance of a data warehouse. Iceberg provides ACID transactions and schema evolution, while Trino enables federated queries across multiple systems without moving data.
R2 serves as the underlying object storage layer, and DataHub ensures that metadata is discoverable and governed. Together, these components allow Cloudflare to enforce data access controls and maintain audit trails for compliance.
Natural Language Access via Skipper
Skipper, the AI analytics agent developed by Cloudflare, allows users to query data using natural language rather than writing SQL directly. This capability lowers the barrier for non technical teams to access billing, security, and business analytics.
Skipper translates natural language requests into SQL queries that run against the Town Lake platform. It also provides explanations of query results in plain language, making the platform accessible to analysts and decision makers who may not have deep database expertise.
The agent is designed to unify access to operational, billing, security, and business data, reducing the need for users to navigate multiple dashboards or data sources.
Billing Queries as a Use Case
The 53 percent share of billing queries highlights the importance of billing analytics for Cloudflare’s internal operations. Billing data includes usage metrics, invoice records, and payment history, which require timely and accurate querying to support customer support, finance, and product teams.
Processing roughly 91,000 billing queries indicates that the platform is handling a significant volume of requests, suggesting it is used regularly for operational reporting and analysis.
The platform’s ability to handle such volumes while maintaining governance suggests that the lakehouse architecture is scalable for enterprise grade workloads.
Implications for Data Governance and Cross System Analytics
Cloudflare emphasized that Town Lake enables governed cross system analytics, meaning data from different departments and storage backends can be queried together under a unified access control framework. This reduces data silos and allows for holistic analysis across operational and business metrics.
The platform integrates with Cloudflare’s existing security and compliance controls, ensuring that sensitive billing and security data remains protected even when accessed through natural language queries.
By using DataHub for metadata management, Cloudflare can track data lineage, schema changes, and usage patterns, which aids in troubleshooting and optimizing query performance over time.
The company has not announced any plans to make Town Lake or Skipper available as commercial products. The disclosure appears to be an internal case study shared to demonstrate how Cloudflare applies modern data architecture and AI to its own operations.
Observers expect that learnings from Town Lake and Skipper may influence future product features or integrations across Cloudflare’s suite of services, particularly in analytics and observability.
