Case Study: Numberly achieves 80% operational footprint reduction and cost savings with ScyllaDB

Numberly Moves from MongoDB to Scylla to Simplify Operations and Reduce Costs

Numberly, a programmatic advertising firm that builds real-time bidding, targeted display, CRM and data management solutions, faced rising operational complexity and SLA risk from running real-time ID matching on MongoDB and batch matching on Apache Hive. Maintaining two copies of every ID table, suffering lost write throughput from MongoDB’s primary/secondary architecture, and failing to meet low-latency read/write SLAs motivated Numberly to evaluate alternatives and ultimately choose ScyllaDB (Scylla Enterprise) as the replacement.

After rigorous load testing, Numberly replaced a 15-node MongoDB cluster with a 3-node Scylla cluster (ScyllaDB/Scylla Enterprise). The ScyllaDB deployment delivered measurable results: roughly an 80% reduction in footprint, significant cost savings, consistently low upsert/read latencies, better data consistency, improved production reliability and datacenter awareness, and overall performance boosts.

Numberly

Alexys Jacob-Monier

Chief Technology Officer