Justin J. Meza

I research the design and operation of efficient and reliable data center infrastructure at Meta Platforms. My interest lies in designing and operating efficient and reliable planetary-scale computers, with which to organize unimaginably large collections of hardware, software, and information. I also go by the Chinese name 梅杰士My name seal. My name seal, set using seal script on a stone stamp, and pressed in vermillion pigment..

I received my MS and PhD in Electrical and Computer Engineering at Carnegie Mellon University and my BS in Computer Science at University of California at Los Angeles. You may contact me at , although I do not respond to recruiting messages.

Research and Development

I develop tools and techniques to improve the reliability of services used by billions of people. I am most interested in understanding fundamental principals to safeguard against the emergent behaviorFor a survey of emergent behavior in distributed systems, see Jeff Mogul’s excellent discussion, Emergent (Mis) behavior vs. Complex Software Systems. of interconnected and geographically-distributed populations of services. The main research question I seek to answer is,

“How can we treat an entire planet as a single ‘computer?’”

Here, I summarize the artifacts of my work in the following areas:

Data Center Infrastructure

“Taiji: Managing Global User Traffic for Large-Scale Internet Services at the Edge,” by David Chou, Tianyin Xu, Kaushik Veeraraghavan, Andrew Newell, Sonia Margulis, Lin Xiao, Pol Mauri Ruiz, Justin Meza, Kiryong Ha, Shruti Padmanabha, Kevin Cole, and Dmitri Perelman in Symposium on Operating Systems Principles, 2019.

“Maelstrom: Mitigating Datacenter-Level Disasters by Draining Interdependent Traffic Safely and Efficiently,” by Kaushik Veeraraghavan, Justin Meza, Scott Michelson, Sankaralingam Panneerselvam, Alex Gyori, David Chou, Sonia Margulis, Daniel Obenshain, Shruti Padmanabha, Ashish Shah, Yee Jiun Song, and Tianyin Xu in USENIX Symposium on Operating Systems Design and Implementation, 2018.

“A Large Scale Study of Data Center Network Reliability,” by Justin Meza, Tianyin Xu, Kaushik Veeraraghavan, and Onur Mutlu in ACM Internet Measurement Conference, 2018.

“Kraken: Leveraging Live Traffic Tests to Identify and Resolve Resource Utilization Bottlenecks in Large Scale Web Services,” by Kaushik Veeraraghavan, Justin Meza, David Chou, Wonho Kim, Sonia Margulis, Scott Michelson, Rajesh Nishtala, Daniel Obenshain, Dmitri Perelman, and Yee Jiun Song in USENIX Symposium on Operating Systems Design and Implementation, 2016.

“Dynamo: Facebook’s Data Center-Wide Power Management System,” by Qiang Wu, Qingyuan Deng, Lakshmi Ganesh, Chang-Hong Raymond Hsu, Yun Jin, Sanjeev Kumar, Bin Li, Justin Meza, and Yee Jiun Song in International Symposium on Computer Architecture, 2016.

“Gorilla: A Fast, Scalable, In-Memory Time Series Database,” by Tuomas Pelkonen, Scott Franklin, Justin Teller, Paul Cavallaro, Qi Huang, Justin Meza, and Kaushik Veeraraghavan in International Conference on Very Large Data Bases, 2015.

“A Large-Scale Study of Flash Memory Errors in the Field,” by Justin Meza, Qiang Wu, Sanjeev Kumar, and Onur Mutlu in ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems, 2015.

“Revisiting Memory Errors in Large-Scale Production Data Centers: Analysis and Modeling of New Trends from the Field,” by Justin Meza, Qiang Wu, Sanjeev Kumar, and Onur Mutlu in IEEE/IFIP International Conference on Dependable Systems and Networks, 2015. (Covered on Hacker News, ZDNet, and The Register.)

Non-Volatile Memory

“The Main Memory System: Challenges and Opportunities,” by Onur Mutlu, Justin Meza, and Lavanya Subramanian in Communications of the Korean Institute of Information Scientists and Engineers, 2015. (Invited article.)

“Efficient Data Mapping and Buffering Techniques for Multi-Level Cell Phase-Change Memories,” by HanBin Yoon, Justin Meza, Naveen Muralimanohar, Norman P. Jouppi, and Onur Mutlu in ACM Transactions on Architecture and Code Optimizations, 2014. (Best presentation award at HiPEAC 2015.)

“Characterizing Application Memory Error Vulnerability to Optimize Data Center Cost via Heterogeneous-Reliability Memory,” by Yixin Luo, Sriram Govindan, Bikash Sharma, Mark Santaniello, Justin Meza, Aman Kansal, Jie Liu, Badriddine Khessib, Kushagra Vaid, and Onur Mutlu in IEEE/IFIP International Conference on Dependable Systems and Networks, 2014. (Covered on ZDNet.)

“A Case for Efficient Hardware/Software Cooperative Management of Storage and Memory,” by Justin Meza, Yixin Luo, Samira Khan, Jishen Zhao, Yuan Xie, and Onur Mutlu in Workshop on Energy-Efficient Design, 2013.

“Row Buffer Locality-Aware Caching Policies for Hybrid Memories,” by HanBin Yoon, Justin Meza, Rachata Ausavarungnirun, Rachael Harding, and Onur Mutlu in International Conference on Computer Design, 2012. (Best paper award in computer systems and applications track.)

“A Case for Small Row Buffers in Non-Volatile Main Memories,” by Justin Meza, Jing Li, and Onur Mutlu in International Conference on Computer Design, 2012. (Poster session.)

“Evaluating Row Buffer Locality in Future Non-Volatile Main Memories,” by Justin Meza, Jing Li, and Onur Mutlu in SAFARI Technical Report, 2012.

“Enabling Efficient and Scalable Hybrid Memories using Fine-Granularity DRAM Cache Management,” by Justin Meza, Jichuan Chang, HanBin Yoon, Onur Mutlu, and Parthasarathy Ranganathan in , 2012.

“Row Buffer Locality-Aware Data Placement in Hybrid Memories,” by HanBin Yoon, Justin Meza, Rachata Ausavarungnirun, Rachael Harding, and Onur Mutlu in SAFARI Technical Report, 2011.

Energy Efficiency and Sustainability

“Totally Green: Evaluating and Designing Servers for Lifecycle Environmental Impact,” by Jichuan Chang, Justin Meza, Parthasarathy Ranganathan, Cullen Bash, and Amip Shah in International Conference on Architectural Support for Programming Languages and Operating Systems, 2012.

“Green Server Design: Beyond Operational Energy to Sustainability,” by Jichuan Chang, Justin Meza, Parthasarathy Ranganathan, Cullen Bash, and Amip Shah in Workshop on Power-Aware Computing and Systems, 2010.

“Lifecycle-Based Data Center Design,” by Justin Meza, Rocky Shih, Amip Shah, Parthasarathy Ranganathan, Jichuan Chang, and Cullen Bash in ASME International Mechanical Engineering Congress and Exposition, 2010.

“Tracking the Power in an Enterprise Decision Support System,” by Justin Meza, Mehul A. Shah, Parthasarathy Ranganathan, Mike Fitzner, and Judson Veazey in International Symposium on Low Power Electronics and Design, 2009.

“Energy Efficiency: The New Holy Grail of Data Management Systems Research,” by Stavros Harizopoulos, Mehul A. Shah, Justin Meza, and Parthasarathy Ranganathan in Conference on Innovative Data Systems, 2009.

“Models and Metrics to Enable Energy-Efficiency Optimizations,” by Suzanne Rivoire, Mehul A. Shah, Parthasarathy Ranganathan, Christos Kozyrakis, and Justin Meza in IEEE Computer, 2007. (Cover feature.)

Books

“Large Scale Studies of Memory, Storage, and Network Failures in a Modern Data Center,” by Justin Meza in ProQuest, 2018. (PhD thesis dissertation.)

“Computer Architecture: A Quantitative Approach,” by John L. Hennessy and David A. Patterson in Elsevier, 2012. (Acknowledgement credit for chapter 6 exercises.)

Patents

Managing an Infrastructure Housing Disaggregated Heat Sources. A way to manage cooling in a datacenter with heterogeneous clusters of resources with different heat intensities.

Computer Racks. Spine-like computer racks containing resource (compute, memory, storage, and so on) blades optimized for lifetime sustainability.

Data Migration between Memory Locations. A policy to efficiently migrate data between different storage devices with different granularities of data access.

Storing Cache Metadata Separately from Integrated Circuit Containing Cache Controller. A technique to store and retrieve cache metadata without using expensive on-controller resources.

⁂

Personal Projects

press: The language, file format, and system I use to typeset my digital manuscripts, including the one you are reading right now. Currently under development.

memories: I am fascinated by what goes on in computer memory and wrote a simple program, memories, to visualizeA memory range visualized by memories. A memory range, as rendered in three dimensions by memories: Each colored cube represents the contents of a 32-bit value at a virtual memory address. memory activity.

lolcode: I develop and maintain the most widely-used interpreter for the lolcode esoteric programming langugage. I now maintain the lolcode specification as well as the lolcode website.

httpd.lol: httpd.lol is a ridiculous bit of code that I wrote to implement a simple HTTP server written in lolcode.

inet: inet is a simple TCP/UDP communication library written in C. It abstracts away the icky socket and protocal interfaces into a more declarative host-based model of communication.

triad: triad is an implementation of the Chord peer-to-peer lookup service in C. It maps 64-bit integers to IP addresses and can be used as the underlying service-discovery layer for a distributed system.

wincurses: wincurses is an ncurses implementation for Windows terminal emulators.

⁂

Last published on Friday, September the 16th, 2022, in Menlo Park, California, USA, by jjm using press build f94f909 on revision cefa4f5.