Case Study: MoPub (Twitter) achieves seconds-scale interactive analytics on terabytes of ad data with Imply

Interactive Analytics at MoPub (Twitter) Using Druid and Imply to Query Terabytes of Data in Seconds

MoPub, a Twitter company that provides monetization for mobile app publishers, faced an extreme scale and observability challenge: their platform addresses about 1.7 billion monthly unique devices, 1 trillion monthly ad requests, 52,000 apps and 180 demand partners. To give customers interactive analytics and rapid troubleshooting, MoPub built MoPub Analytics on Apache Druid and exposed it with Imply Pivot, aiming to query terabytes of data and answer ad‑revenue questions in seconds.

MoPub implemented a Druid-based pipeline with HDFS deep storage, historicals, brokers and Imply Pivot as the drag‑and‑drop UI, ingesting over 9 billion aggregated rows per day from ~30 billion daily ad requests (150 TB of raw logs). The Imply-enabled solution lets users slice and dice many time windows, dimensions and metrics interactively, with queries returning in seconds—enabling faster root‑cause analysis and revenue optimization at massive scale.

MoPub