Case Study: Global Ad Tech Company achieves 5x faster online profile updates with Aerospike

Aerospike and Apache Spark Power a Real-time Online Profile Store at a Global Ad Tech Company

A global ad-tech firm building an Online Profile Store to consolidate current, historical and derived consumer data for ML and BI struggled with an internally built system that caused data-integration and quality issues, an overly complex infrastructure, and prohibitively slow Spark jobs (around 12 hours), leaving the platform unable to meet low-latency, high-scale needs or planned data growth.

The company deployed Aerospike with Aerospike Connect for Spark in the cloud—33-node Aerospike and ~300-node Spark clusters—to enable massive parallel data transfer and processing. The solution processed 13 billion objects (150 TB), mapped 4,096 Aerospike partitions to up to 32,768 Spark partitions (8,192 Spark tasks on ~7,500 cores), reduced the profile-store update job from 12 hours to 2.4 hours (≈10x faster insight), improved data quality, accelerated production rollout to months, and lowered TCO and operational footprint.