Scaling from 2K engineers to 12K engineers and 10K deploys per day

In times of austerity, efficiency concerns are greater in companies that want to continue growing solidly. Everyone discusses the best approaches for scalability, resiliency, governance, and growth. CEOs putting pressure on their teams to reduce expenses, improve processes and deliver more with less are a constant theme in the news. At Mercado Libre, the largest e-commerce, fintech & logistics company in LATAM with over 110 million active users, such concerns are an essential part of our culture, which moved us to seek the best practices since 1999, and with a natural process of searching for efficiency, led us to build our platform, Fury, which has already reached a very high maturity level, handling 900M+ requests per minute using 114K+ instances distributed among different clouds, orchestrating more than 26K microservices. We build a NoOps culture for our 12k engineers, which allows them to focus on creating great products in terms of architecture and functionalities, giving us an excellent time to market, going from zero to production in four steps, with an exceptional ability to react to problems. This talk shares our platform's benefits, the lessons learned on the journey to build it, trade-offs, and mistakes we made. We will perform a general view on Fury and then go deep into some solutions that we have implemented, and that can add value to the audience, such as: Dev Experience: How Fury optimizes Developer's productivity and creates the NoOps culture by providing Out-of-the-box management of code repositories, a robust CI/CD Pipeline, metric providers for complete observability and Cost Management; Opportunities creation that suggests or applies updates for Critical Vulnerabilities, cost optimization, and others; Performing feature inference using Sidecars and assigning better control about Security, Compliance, and NoOps on each of our microservices. Meet some of our sidecars such as Mysql Proxy, Traffic Middleware, Logs, and Metrics Agents; Acting quickly to solve issues such as Availability Zone's outages, the log4j vulnerability, or promptly responding to government regulations.


Speaker

Sebastian Barrios

VP of Technology @Mercado Libre

Sebastian is Mercado Libre’s VP of Technology. He leads all cross-technology teams such as Platform Infrastructure, Services, Architecture, Machine Learning, Data Engineering, SRE, Mobile, Frontend, and UX. As a serial entrepreneur turned executive, Sebastian started and sold multiple companies, the latest being Yaxi, which was acquired by Cabify, where Sebastian served as CTO for 3 years. He was recognized as the youngest ever Endeavor Entrepreneur (he was 22 at the time). A record he still holds today. In 2020 he was recognized on the Forbes “30 under 30” ranking. He graduated with honors with a Bachelor's Degree in Computer Science from ITAM (Instituto Tecnológico Autónoma de México)

Read more

Speaker

Lucia Brizuela

Sr Tech Director @Mercado Libre

Lucia Brizuela is a Sr Tech Director at MercadoLibre leading the Platform and Knowledge Management IT teams, focused on improving the experience and efficiency of the development team from different angles. She has been in MercadoLibre since 2009 and has more than 10 years of experience in staff teams.

Read more

Session Sponsored By

Mercado Libre is the #1 e-commerce technology and financial services company in Latin America

Date

Wednesday Oct 26 / 10:35AM PDT ( 50 minutes )

Share

From the same track

Session

Tailor Your Processes to be as Data-Driven as Your Systems

Wednesday Oct 26 / 11:50AM PDT

Engineering leaders recognize the importance of building observability into their software systems. Through testing, monitoring, and detailed reporting, it’s possible to thoroughly understand what’s working and what isn’t. We always know when our code is working – why not our teams?

James McGill

VP of Engineering @Code Climate

Session

Building Resilient Microservices: The hard parts of a distributed architecture

Wednesday Oct 26 / 02:55PM PDT

Building resilience is a complex theme in any distributed architecture; even for modern cloud-native applications, resources could be temporal, the network is unreliable, and there are always communication challenges between services and broken dependencies.

Wanderlei Souza

Distinguished Engineer @Build by McKinsey