Caesar contributes to Internet infrastructure research
One of the CSL’s newest assistant research professors, Matthew Caesar, may be young, but he has already accomplished milestones -- and major companies are taking notice. A system he created has been implemented on one of the largest ISPs in the world. In addition, he targets his research at major network operators such as AT&T and Sprint, equipment vendors such as Cisco, and enterprise network operators such as Microsoft.
"Matt Caesar brings much needed expertise on Internet infrastructure to CSL,” said CSL Assistant Research Professor Nikita Borisov. “His research has connections to many CSL core research areas, including reliability, dependability, security and networking. Prof. Caesar has been a great collaborator and a wonderful second mentor to several students in my group."
Caesar primarily researches and teaches computer networks and systems.
“Network software today is extremely complex, comprised of millions of lines of code and distributed across vast numbers of routers and servers,” Caesar said. “Most of the complexity of modern data centers, and wide-area and enterprise networks lies in this software. I focus on making this software robust to malicious behavior and errors.”
Although involved in many projects, two particularly innovative endeavors stand out.
First, Caesar is building systems that can automatically recover from errors. In the paper “Building Bug-Tolerant Routers with Virtualization,” he proposed the use of a bug-tolerant router to run multiple diverse copies of router software concurrently. In the case that one copy fails, there would be a back-up copy to automatically recover the error.
Caesar compares this system to redundant brake lines on modern cars.
“The idea is if you just had one set of brake lines, and it failed, then you’d crash. But if you have two sets of brake lines, then they’d both have to fail for your car to crash, and the chance of that is much lower than just one of the lines failing,” he explained. “We’re doing the same thing with software: instead of running just one copy of software, we run three copies. All three copies have to fail before our system goes down."
Second, Caesar has created a system that may make networked software easier to debug.
“When problems arise on computers, humans need to debug software, which is hard today because systems are so complex,” Caesar said. “We’re working on a system that makes it easier for people to go in and successfully debug networks.”
He hopes this research will help major companies save revenue.
“Every once in awhile you’ll read about how a major company had an outage or lost data,” Caesar said. “When you have a large ISP with a bug in its system that causes a large website to go offline, millions of dollars could be lost.”
Caesar received his Ph.D. in Computer Science from UC Berkeley in 2007, and Illinois hired him the following year to teach undergraduate and graduate level courses in computer science, including CS 438 (Communication Networks), CS 498 (Systems and Networking Laboratory) and CS 598 (Advanced Internetworking). He joined the Coordinated Science Laboratory in Spring 2009.
“So far, my experience at the CSL has been rewarding in several ways,” he said. “I like the chance to interact with other professors on a variety of projects. I also enjoy the multi-disciplinary focus – it allows me to interact and learn from faculty and students in other areas.”
Caesar is a 2010 AT&T VURI Awardee and Member of the DARPA 2010 Computer Science Study Group.