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

by Rick Bilodeau · July 5, 2019

MoPub, a Twitter company, provides monetization solutions for mobile app publishers and developers around the globe. They operate at a massive scale. The MoPub platform addresses over 1.7 billion monthly unique devices, 1 trillion monthly ad requests, 52,000 apps, and 180 demand side partners.

MoPub has just launched a new solution called MoPub Analytics based on the Apache Druid real-time analytics database and using Imply Pivot as the drag-and-drop UI. The solution allows MoPub customers to answer business questions, troubleshoot issues and optimize revenue. Users can determine the root cause of new data trends by interactively analyzing the data across many different time slices, dimensions, and metrics.

Every day, MoPub runs over 30 billion ad requests, which generate over 150 terabytes of raw logs. Even after aggregation they ingest over 9 billion rows into the analytics engine each day. Queries on MoPub Analytics return in seconds.

You can find the MoPub Analytics post on the Twitter Engineering blog.

MoPub Analytics Architecture

architecture-diagram-mopub-analytics-twitter

Back to blog

How can we help?