Case Study: Eniro achieves low-latency, cost-effective high-performance data streaming with ScyllaDB

Eniro Chooses Scylla over Redis and Kafka for Data Streaming Application

Eniro, a leading Nordic local-search company serving 23.2 million unique monthly visitors and 70.6 million searches, faced performance limits from a single-threaded Redis setup, high read/write latencies, and the need for a persistent, multi-type data store that could integrate with streaming and analytics tools. To address these challenges they chose ScyllaDB as the prime data store (using NATS for event streaming) to replace Redis/Kafka and remove end-application resource constraints and service stalls.

ScyllaDB was deployed as the primary datastore with application logic in Go (gocql) and NATS handling website events, logs and documents; Scylla stored nightly artifacts, online and real-time data across keyspaces. The deployment delivered a durable, cost-effective streaming solution: read latencies fell to single-digit milliseconds, write latencies to sub-milliseconds (with operators observing microsecond-level responses), supported multiple data types in one cluster, ran efficiently on three servers (16 GB RAM, 6-core Intel Xeon), and provided seamless Go support and low-effort upgrades with ScyllaDB.

Eniro

Henrik Johansson

Senior Developer