Case Study: FactSet achieves sub-20ms query performance with Dgraph

160 Million Nodes and Two Billion Edges: How FactSet Uses Dgraph and GraphQL+- to Power One of The Largest Financial Databases in the World

FactSet, a global provider of financial information and analytics, needed to modernize its Core Symbology database, which tracks billions of financial objects across dozens of SQL and NoSQL systems. As its schema grew more complex, the company faced difficult joins, high query latency, and limited usability for internal developers. FactSet chose Dgraph and its GraphQL+- query interface to create a more flexible, distributed graph-based foundation for its symbol data.

With Dgraph, FactSet refactored Core Symbology into a single cluster with 160 million nodes and 2 billion edges, using facets to support dual time-series data and versioning. The result was a simpler schema, faster development for nontechnical users, and query performance under 20 milliseconds for most requests, all while reducing infrastructure footprint and improving database management. FactSet also reported high reliability during testing, with no downtime and quick recovery from cold starts.

FactSet

Mark Boxall

Principal Software Engineer