06_BULLET: LOOK-FORWARD QUERYING

1 .A REAL TIME DATA QUERY ENGINE Akshai Sarma, Nathan Speidel, Michael Natkovich 

2 . MOTIVATION: INSTRUMENTATION & THE CYCLE OF SADNESS • Code on web pages/apps to capture usage and behavior • Drives all data applications • But validation is unbearably slow • Needs to be seconds not minutes/hours • Needs to be easy to query • Needs programmatic access 2 

3 . A TYPICAL WAY OF QUERYING 3 

4 . ATYPICAL WAY OF QUERYING 4 

5 . BULLET: LOOK-FORWARD QUERYING • No persistence: light, fast and scalable! • UI and API for ad-hoc or programmatic usage • Pluggable with your streaming data • Platform agnostic • Queries run for time windows and can: • Filter data: Equals, Regex, And, Or and more • Get Raw or Aggregate data: Group By, Count Distinct, Top K, and more 5 

6 . DATASKETCHES • Sketches are a class of stochastic streaming algorithms • Provides approximate results (if data is too large) • Provable error bounds • Fixed memory footprint • Mergeable, allowing for parallel processing 6 

7 . ARCHITECTURE 7 

8 .8 

9 . FULLY OPEN-SOURCED • We are on GitHub - contributions, ideas, feedback welcome! Component Repo Storm https://github.com/yahoo/bullet-storm Core https://github.com/yahoo/bullet-core API https://github.com/yahoo/bullet-service UI https://github.com/yahoo/bullet-ui Record https://github.com/yahoo/bullet-record Kafka https://github.com/yahoo/bullet-kafka 9 

10 . LINKS • Contact us • Developers : bullet-dev@googlegroups.com • Users : bullet-users@googlegroups.com • For all Documentation: Usage, Performance, Quick Start, Examples • https://yahoo.github.io/bullet-docs • DataSketches: https://datasketches.github.io 10