🤖Uber's SubmitQueue Cuts CI Resource Usage by 53%
Uber slashes CI costs with machine learning
TL;DR
Uber's SubmitQueue uses ML to predict build success and execution times, slashing CI resource usage by 53% and speeding up PR landing times by 37%. Worth watching for teams with large monorepos.
Uber's SubmitQueue slashes CI resource usage by 53% and accelerates PR landing times by 37% through machine learning and binary speculation trees. This is a big deal for teams managing large monorepos with thousands of monthly changes. SubmitQueue's conflict analysis and build prediction models ensure efficient use of CI resources, critical for maintaining productivity in massive development environments. Uber's monorepos handle 65,000+ monthly changes and 135 million lines of code, making efficient CI management essential.
Key Points
SubmitQueue uses binary speculation trees and machine learning to predict build success and execution times.
CI resource usage slashed by 53% while PR landing times improved by 37%.
Handles 65,000+ monthly changes and 135 million lines of code across 6 monorepos.
Facilitates 100,000+ deployments per month, critical for Uber's global development.
SubmitQueue's conflict analysis determines the probability of build needed for each node.
Why It Matters
If you're managing a large monorepo with thousands of monthly changes, SubmitQueue's ML-driven approach to CI optimization could save you a fortune. For Uber, it means 53% less CI resource usage and 37% faster PR landing times. But the real win is for teams with similar scale and complexity, where efficient CI management is crucial for maintaining productivity and avoiding red mainlines.
Comments
Be the first to comment
Enjoyed this article?
Get it daily. 7am. Free. Reads in 5 minutes.
Join 3,556 builders reading daily.