Case Study: Zeotap Builds a 20 Billion ID Graph with ScyllaDB and JanusGraph

Zeotap A Graph of Twenty Billion IDs Built on ScyllaDB and JanusGraph

Zeotap is a customer intelligence platform that needed to resolve and correlate massive volumes of identity data across devices and channels while meeting strict privacy, compliance, and freshness SLAs. As its original Spark-based batch architecture grew past 3 billion IDs, processing times stretched beyond 24 hours, making it hard to support high read/write throughput, TTL, and fast match/export queries.

ScyllaDB, paired with JanusGraph, gave Zeotap a scalable graph database foundation for its Connect product, with low-latency neighborhood traversal, high-ingestion throughput, and first-class linkage metadata. By moving to JanusGraph on ScyllaDB, Zeotap built what it described as the world’s largest native JanusGraph database, now tracking over 20 billion IDs, while improving query performance, supporting hundreds of millions of new data points per day, and reducing production costs compared with memory-heavy alternatives.

Zeotap

Sathish K S

Vice President Engineering