- Tech lead for the Orion Routing Engine, Google’s software-defined network controller that computes data-center routing in the abstract and programs the fleet of network switches.
Sunghwan Yoo
Staff Software Engineer at Google · Sunnyvale, California
Software-defined networking for Google’s data centers.
About
I am a staff software engineer at Google Cloud on the Network Infrastructure team, where I am tech lead for the Orion Routing Engine — Google’s software-defined network controller that computes data-center routing in the abstract and programs the switches accordingly. For most of my career I have designed, built and operated SDN control software at that scale.
My technical interests are networking, routing, distributed systems and cloud infrastructure, alongside the craft of software engineering itself. I also care about building and leading good teams, and I enjoy strategically complex problems whether they are technical or organizational.
My Ph.D. work at Purdue was on networking, distributed systems, parallel computing and fault tolerance. Outside work I read a great deal — around 170 books a year — play unhurried games (RPGs, SRPGs, puzzles), and keep a few hobby projects going.
Experience
- SDN controller software running in Google data centers: design, implementation and test.
- Aquila: a unified, low-latency fabric for datacenter networks (NSDI ’22).
- SDN controller software running in Google data centers: design, implementation and test.
- Mesa, the large-scale distributed database behind Google’s Ads infrastructure.
Purdue University
- Problem Solving & Object-Oriented Programming (CS 180); Computer Networks (CS 422).
- EventWave — a scalable, elastic distributed event-driven framework (SoCC ’13).
- MaceKen — a fault-tolerant distributed systems framework (USENIX ATC ’12).
- InContext — annotation-based multi-threading for event-driven systems (HPDC ’11).
- MapReduce in Mace — a high-performance distributed computing application (HPDC ’11).
- GridSphere-based web service serving LARS, GOES, NEXRAD and climate radar data, backed by SRB/iRODS data management.
- Extreme Collaboration: a highly available distributed computation model for biological problems.
- Finding correlated evolution patterns in DNA and protein sequences.
Hewlett-Packard Laboratories
- Early version of MaceKen, a fault-tolerant distributed systems framework.
Yonsei University
- Web-based OLAP application in PHP, Java and MySQL.
- Java user interface for a statistical intrusion-prevention system.
- Administration of the lab’s machines.
- Automata and Formal Languages.
Education
Dissertation: a unified model for building multi-threaded, fault-tolerant distributed systems frameworks.