When Segment Met Cricket: A Scaling Story

Thursday, 30 August, 2018 - 14:4515:30

Michael Fischer, Segment

Abstract: 

Segment receives over a half-million events per second from web sites, mobile applications, and other data sources around the world. Last summer, an unexpectedly popular event pushed our AWS-and-Kafka-based event pipeline to its limits and tested our ability to scale quickly to handle the crushing demand. Yet through access to crucial data and quick thinking, we were able to avoid catastrophe. We’ll discuss in detail the events that led to the crisis, assumptions we made that proved to be false, the key observations we had that helped us rescue the system before it was too late, and how we gained a tremendous increase in capacity for a small amount of additional investment.

Michael Fischer, Segment

Michael Fischer is a Site Reliability Engineering lead at Segment, which provides infrastructure for customer data. Prior to Segment, Michael worked at companies including Zendesk and Yahoo! in principal engineering roles. He holds a law degree from Santa Clara University and a BA from the University of California, San Diego. He lives in Oakland, California with his wife and pets; and enjoys scuba diving, concerts, cooking, and world travel.

BibTeX
@inproceedings {218859,
author = {Michael Fischer},
title = {When Segment Met Cricket: A Scaling Story},
booktitle = {SREcon18 Europe/Middle East/Africa (SREcon18 Europe)},
year = {2018},
address = {Dusseldorf},
url = {https://www.usenix.org/node/218860},
publisher = {USENIX Association},
month = aug
}