Instagram is a feature-rich app with over 1 billion monthly active users. The Instagram infrastructure is comprised of hundreds of servers running in different geographic locations and hosts a multitude of services such as a Python-based frontend, Cassandra key-value stores, ML ranking services, etc. Guilin Chen and Shobhit Kanaujia discuss infrastructure practices for supporting various initiatives for enriching user experience, for example, but not limited to, the launch of new products and features. The objective of the various efficiency initiatives under play is to minimize the total number of servers used by all of the services that support Instagram.
Guilin and Shobhit walk you through their experiences in regard to meeting the efficiency goals at Instagram. In particular they discuss: How do they arrive at their server-capacity needs—two inputs to this process are organic user-growth projection and a regression allowance for each of the services; how to define monitoring metrics (weights) to ensure services are operating within their regression allowance—these metrics help determine success or failure of meeting the efficiency goal; how do they approach regression detection—they also share a snapshot of some of the techniques they have employed to address a subset of these regressions; and how to simulate DR scenarios to expose issues arising due to interdependency among services?
Guilin Chen is a software engineer at Facebook, working on mobile performance, and leads the Instagram efficiency team.
Shobhit Kanaujia is an engineer at Facebook who specializes in full stack performance and efficiency at scale.
Help us make this conference the best it can be for you. Have questions you'd like this speaker to address? Suggestions for issues that deserve extra attention? Feedback that you'd like to share with the speaker and other attendees?
Join the conversation here (requires login)
©2019, O'Reilly Media, Inc. • (800) 889-8969 or (707) 827-7019 • Monday-Friday 7:30am-5pm PT • All trademarks and registered trademarks appearing on oreilly.com are the property of their respective owners. • email@example.com