Case Study: Zenly achieves real-time sub-ms latency and 100% uptime with ScyllaDB

Zenly Pinpoints Scylla to Bring High Performance and Low Latency to Its Popular Geo-Location App

Zenly, a Paris-based real-time geo‑location app (now part of Snap, Inc.), needed a faster, simpler data platform to support rapid, frequent location updates for over a million users. Its prior stack—Elasticsearch, massive Redis caches, and Cassandra—was complex to operate and introduced latency and operational overhead, so Zenly sought a unified low‑latency database and evaluated ScyllaDB’s Scylla as an alternative.

Zenly migrated to ScyllaDB’s Scylla across four clusters (two single‑region, two double‑region) and removed the need for external caching, simplifying operations and isolating features for redundancy. The ScyllaDB deployment delivered measurable gains: 100% uptime across the clusters, in‑production throughput of 300K requests on seven nodes, p95 latency <1 ms and p99 latency <3 ms, enabling Zenly to stay “live always” with much higher performance.

Zenly

Jean-Baptiste Dalido

Head of Infrastructure