July 20–24, 2015
Portland, OR

Designing for failure: How Uber scaled its realtime market platform

Matt Ranney (Uber)
2:30pm–3:10pm Wednesday, 07/22/2015
Architecture Portland 256
Average rating: ****.
(4.70, 23 ratings)
Slides:   external link

Prerequisite Knowledge

This talk will be most useful for people who are working on a growing service, or who want to better prepare for how to scale their service. If you've ever grappled with the difficult tradeoffs of eliminating single points of failure, then this talk is for you.


As Uber’s business grows, more engineers are added, more features are requested, and expectations rise. We’ve redesigned the back end to handle these scaling issues, with a focus on handling failures.

We’ll describe the architecture we’ve come up with, and how it behaves when things break. We’ll also cover some building blocks of this system that we’ve open sourced.

Photo of Matt Ranney

Matt Ranney


Matt Ranney is chief systems architect for Uber. Before that, he was cofounder and CTO of Voxer.

Comments on this page are now closed.


Jieun Lee
07/23/2015 8:55am PDT

It was great presentation. Thank you

Srivatsa Radhakrishna
07/22/2015 10:02am PDT

Hi Matt,

Great talk, can you please share your slides?