The Distributed System Research Group at University of Toronto focuses on improving the reliability, failure diagnosis, and performance of real systems software.
[Impact] Yongle’s PhD dissertation won The SIGOPS Dennis M. Ritchie Thesis Award. Congrats Yongle!
[Impact] Received Meta 2022 Systems Research Award for CLP. Thanks Meta!
[Startup] CLP is deployed on Uber’s big data platform. In 2021, Uber’s growth eclipsed the capability of their existing log management tools, forcing them to start dropping Petabytes of INFO-level Spark logs due to scalability issues (e.g., capacity, SSD burn-out). We integrated CLP into their logging library (Log4j) on Uber’s big data platform, achieving a 169x compression ratio. Now, all logs are retained, and efficiently analyzed, at 169x less cost. Check out Uber’s Engineering Blog for more details. (Also checkout the Hacker News discussion.)
[Publication] Our ATC'22 paper on runtime performance is invited to publish in USENIX ;login:.
[Impact] HBASE-12187 and all its subtasks have been closed and implemented. HBase developers opened HBASE-12187 to address the issues we found in our OSDI'14 paper. It consists of 9 subtasks, including integrating Aspirator checks (the tool from our paper) as a check-in policy, use other static checkers including Coverity, thorough code review, etc.
[Publication] Our ctFS paper is invited to publish in USENIX ;login: and ACM TOS by FAST'22 PC.
[Startup] YScope, our startup to work on our open-source log compression and search technology, is launched.
[Impact] Aspirator is now part of Google’s error-prone static checker. See the pull request.