Case Study: Tantan achieves real-time, low-latency matchmaking and lower infrastructure costs with ScyllaDB

Tantan Connects with Scylla to Help Singles Find Their Closest Match

Tantan is a locality-based dating app with more than 160 million users and over 10 million daily active users that must identify nearby matches in real time (about 300 meters). At peak the app sees roughly 50,000 writes per second, and Tantan struggled with latency and spikes while running Cassandra (two 16‑node clusters) due to garbage collection and maintenance tasks. To address this they evaluated ScyllaDB and its Scylla NoSQL database.

Tantan migrated the passby workload to an 8‑node Scylla cluster provided by ScyllaDB, cutting the node count for that function roughly in half and delivering significant cost savings. By adopting ScyllaDB they eliminated garbage‑collection–related latency spikes and other repair-induced delays, achieved the low-latency mixed read/write performance their real‑time matching requires, and reduced operational overhead—with ScyllaDB’s team and community helping optimize the deployment.

Tantan

Ken Peng

Backend Specialist