Case Study: Forensiq achieves 5ms pre-bid fraud detection at 1 trillion bids/month with Aerospike

Forensiq’s Online Ad Fraud Detection Processing 1 Trillion Bid Requests Per Month

Forensiq is a fraud-detection company that filters fraudulent ad impressions in real time, processing about 1 trillion bid requests per month and operating under strict 150 ms RTB constraints that require 5 ms pre-bid decisions. Facing explosive (1,000x) growth and the high stakes of a market where an estimated 36% of impressions are fraudulent, Forensiq needed a database that could deliver extreme speed, reliability and seamless scaling.

Forensiq chose Aerospike’s in-memory NoSQL database over first-generation NoSQL and Redis, deploying it on bare-metal clusters with cross-data-center replication and a custom data-management layer for easy migration. The result: consistent 5 ms SLAs, elimination of caching/sharding complexity, the ability to scale with demand, and more focus on improving reputation-scoring algorithms to boost ad performance and ROI while combating online ad fraud.

Forensiq

Matt Vella

Chief Technology Officer