Today's largest data processing workloads are hosted in cloud data centers. Due to unprecedented data growth, these workloads have ballooned to hyperscale level, encompassing billions to trillions of data items and hundreds to thousands of machines per query. Enabling and expanding with these workloads are highly scalable data center networks that connect up to hundreds of thousands of networked servers. At hyperscale, the classic layered designs are no longer sustainable: without knowing how their data is transferred in the network, applications can make egregious decisions in executions; the cloud infrastructure also performs poorly without rethinking the interfaces and services exposed to its applications. Rather than optimize these massive layers in silos, I build systems across them with principled network-centric designs for efficient hyperscale data processing.

• In current networks, we design data processing systems by adopting network awareness. Considering data center network characteristics in optimizing distributed operations effectively reduces communication cost at hyperscale. We built GraphRex [SIGMOD '19], the first system that systematically adopts network awareness for large-scale graph processing. We also proposed TeShu [CIDR '23], a shuffle layer that encompasses network-aware optimizations and offers them to a variety of distributed data processing systems with templates.
• For future networks, we improve the interfaces and services that the cloud infrastructure offers to applications and codesign data processing systems to achieve optimal performance. In particular, we have been pioneering the work on data processing in disaggregated data centers (DDCs), a promising new cloud proposal. We opened the topic by rethinking data processing system designs in DDCs [CIDR '20] and, for the first time, investigated the effect of DDCs on production systems [VLDB '20], which opened a broad space of research opportunities. We proposed TELEPORT [SIGMOD '22], a novel compute pushdown feature of DDCs for efficient data processing. To realize DDC benefits in today’s clouds, I developed Redy [VLDB '22] with industry. More recently, we applied disaggregated designs to improve resource efficiency for blockchain workloads with FlexChain [VLDB '23] and proposed Cowbird [SIGCOMM '23] to reduce CPU cost in disaggregating memory.
• To facilitate network innovation, we must evaluate the performance of network proposals at scale. This, however, is intractable because of network size and complexity. To support our work and that of others, we developed MimicNet [SIGCOMM '21], the first scalable system that provides accurate performance estimates for data center networks. It models network behavior in FatTree topologies with machine learning techniques and composes large-scale simulations with intra- and inter-cluster approximations.

I have also worked on other aspects of data processing, including data science application with machine learning [CIKM '17], and data processing cost efficiency [SoCC '17]. Recently, I proposed CompuCache [CIDR '22], a new cloud service that exploits spot VMs for remote caching and memory-intensive compute offloading.

* students I advise

DPDPU: Data Processing with DPUs [arXiv]
Jiasheng Hu*, Philip Bernstein, Jialin Li, Qizhen Zhang
Conference on Innovative Data Systems Research, CIDR 2025, to appear

DDS: DPU-optimized Disaggregated Storage
Qizhen Zhang, Philip Bernstein, Badrish Chandramouli, Jiasheng Hu*, Yiming Zheng*
International Conference on Very Large Data Bases, VLDB 2024

Computing in the Era of Large Generative Models: From Cloud-Native to AI-Native
Yao Lu, Song Bian, Lequn Chen, Yongjun He, Yulong Hui, Matthew Lentz, Beibin Li, Fei Liu, Jialin Li, Qi Liu, Rui Liu, Xiaoxuan Liu, Lin Ma, Kexin Rong, Jianguo Wang, Yingjun Wu, Yongji Wu, Huanchen Zhang, Minjia Zhang, Qizhen Zhang, Tianyi Zhou, Danyang Zhuo
CoRR, arXiv:2401.12230, 2024

Cowbird: Freeing CPUs to Compute by Offloading the Disaggregation of Memory
Xinyi Chen, Liangcheng Yu, Vincent Liu, Qizhen Zhang
Annual Conference of the ACM Special Interest Group on Data Communication, SIGCOMM 2023

Disaggregated Database Systems
Jianguo Wang, Qizhen Zhang
ACM International Conference on Management of Data, SIGMOD 2023 Tutorial

FlexChain: An Elastic Disaggregated Blockchain
Chenyuan Wu, Mohammad Javad Amiri, Jared Asch, Heena Nagda, Qizhen Zhang, Boon Thau Loo
International Conference on Very Large Data Bases, VLDB 2023

Templating Shuffles
Qizhen Zhang, Jiacheng Wu, Ang Chen, Vincent Liu, Boon Thau Loo
Conference on Innovative Data Systems Research, CIDR 2023

Optimizing Data-intensive Systems in Disaggregated Data Centers with TELEPORT
Qizhen Zhang, Xinyi Chen, Sidharth Sankhe, Zhilei Zheng, Ke Zhong, Sebastian Angel, Ang Chen, Vincent Liu, Boon Thau Loo
ACM International Conference on Management of Data, SIGMOD 2022

Redy: Remote Dynamic Memory Cache
Qizhen Zhang, Philip Bernstein, Daniel Berger, Badrish Chandramouli
International Conference on Very Large Data Bases, VLDB 2022

CompuCache: Remote Computable Caching using Spot VMs
Qizhen Zhang, Philip Bernstein, Daniel Berger, Badrish Chandramouli, Vincent Liu, Boon Thau Loo
Conference on Innovative Data Systems Research, CIDR 2022

MimicNet: Fast Performance Estimates for Data Center Networks with Machine Learning
Qizhen Zhang, Kelvin K.W. Ng, Charles W. Kazer, Shen Yan, João Sedoc, Vincent Liu
Annual Conference of the ACM Special Interest Group on Data Communication, SIGCOMM 2021

Understanding the Effect of Data Center Resource Disaggregation on Production DBMSs
Qizhen Zhang, Yifan Cai, Xinyi Chen, Sebastian Angel, Ang Chen, Vincent Liu, Boon Thau Loo
International Conference on Very Large Data Bases, VLDB 2020

Rethinking Data Management Systems for Disaggregated Data Centers
Qizhen Zhang, Yifan Cai, Sebastian Angel, Ang Chen, Vincent Liu, Boon Thau Loo
Conference on Innovative Data Systems Research, CIDR 2020

Optimizing Declarative Graph Queries at Large Scale
Qizhen Zhang, Akash Acharya, Hongzhi Chen, Simran Arora, Ang Chen, Vincent Liu, Boon Thau Loo
ACM International Conference on Management of Data, SIGMOD 2019

Predicting Startup Crowdfunding Success through Longitudinal Social Engagement Analysis
Qizhen Zhang, Tengyuan Ye, Meryem Essaidi, Shivani Agarwal, Vincent Liu, Boon Thau Loo
ACM International Conference on Information and Knowledge Management, CIKM 2017

Architectural Implications on the Performance and Cost of Graph Analytics Systems
Qizhen Zhang, Hongzhi Chen, Da Yan, James Cheng, Boon Thau Loo, Purushotham Bangalore
ACM Symposium on Cloud Computing, SoCC 2017

Program Committee: SIGMOD 2025, VLDB 2025, ICDE 2025, SIGMOD 2024, EuroSys 2024, EDBT 2024, SoCC 2023, CIKM 2023, ICDE 2023, SoCC 2022, CIKM 2022

Session Chair: VLDB 2023, Northwest Database Society (NWDS) Annual Meeting 2023, SoCC 2022, SIGMOD 2022

Journal Reviewer:
IEEE/ACM Transactions on Networking 2023
IEEE Transactions on Knowledge and Data Engineering 2023
IEEE/ACM Transactions on Networking 2022
IEEE Transactions on Parallel and Distributed Systems 2022
IEEE Transactions on Knowledge and Data Engineering 2022

Organizer:
FDL Reading Group, University of Toronto, Fall 2024
DSL Seminar, University of Pennsylvania, Spring 2017 and Fall 2017

Offloading the Tax of Disaggregation
Invited talk, National University of Singapore, Singapore, March 2024

Memory-disaggregated Database Systems
SIGMOD 2023, Seattle, Washington, United States, June 2023

Templating Shuffles
Northwest Database Society (NWDS) Annual Meeting 2023, Redmond, Washington, United States, May 2023

Redy: Remote Dynamic Memory Cache
VLDB 2022, Virtual, September 2022

Optimizing Data-intensive Systems in Disaggregated Data Centers with TELEPORT
SIGMOD 2022, Philadelphia, Pennsylvania, United States, June 2022

Hyperscale Data Processing with Network-centric Designs
Job talk, February - April 2022
CMU, HKUST, NUS, Ohio State U., Simon Fraser U., UC Irvine, UIUC, U. Minnesota Twin Cities, U. Toronto, U. Virginia, U. Waterloo

CompuCache: Remote Computable Caching using Spot VMs
CIDR 2022, Virtual, January 2022

MimicNet: Fast Performance Estimates for Data Center Networks with Machine Learning
SIGCOMM 2021, Virtual, August 2021

Understanding the Effect of Data Center Resource Disaggregation on Production DBMSs
VLDB 2020, Virtual, September 2020
Microsoft Research, Virtual, August 2020

Rethinking Data Management Systems for Disaggregated Data Centers
CIDR 2020, Amsterdam, Netherlands, January 2020

Optimizing Declarative Graph Queries at Large Scale
Microsoft Research, Redmond, Washington, United States, August 2019
SIGMOD 2019, Amsterdam, Netherlands, July 2019

Predicting Startup Crowdfunding Success through Longitudinal Social Engagement Analysis
CIKM 2017, Singapore, November 2017

Architectural Implications on the Performance and Cost of Graph Analytics Systems
SoCC 2017, Santa Clara, California, United States, September 2017