Department of Computer Science, Cornell University.
My research interests are in systems and networking. I am also interested in theoretical problems arising out of building practical systems. My research thus spans (and integrates) systems, networks, and theory.
ragarwal at cs cornell edu (please read this before sending me an email)
CV | Google scholar | Github | Short Bio
- Other recent-ish news
-  Congratulations to Qizhe Cai for winning 2022 Facebook/Meta PhD Fellowship!
-  James and Mary Tien Teaching Award, the highest teaching award in Cornell College of Engineering.
-  Sloan Research Fellowship!
-  NSF CAREER award!
-  $1M NSF award for work on Pancake!
-  Pancake wins Distinguished Paper award at Usenix Security'20!
-  Congratulations to Saksham Agarwal for winning 2019 Google PhD Fellowship!
-  Sincronia wins Best Student Paper award at SIGCOMM'18!
-  $3M NSF award for work on Resource Disaggregation!
||[Cornell Fellowship] [SIGCOMM Best Student Paper Award (Sincronia)] [Google PhD Fellowship] [Outstanding TA award]
||[Meta PhD Fellowship]
||[Cornell Fellowship] [2x Outstanding TA award]
|Sujaya Maiyya (Postdoc, 2022)  → Assistant Professor, University of Waterloo
|Mina Tahmasbi Arashloo (Postdoc, 2020-2022)  → Assistant Professor, University of Waterloo
|Jaehyun Hwang (Postdoc, 2019-2021)  → Assistant Professor, Sungkyunkwan University
|Anurag Khandelwal (Postdoc, 2019)  → Assistant Professor, Yale University
|Katie Gioioso (MS, 2021)  → PhD student, Stanford
- Resource Disaggregation: Shared-nothing architectures provide good data locality and cross-job isolation. However, for modern workloads where peak resource demands can be much higher than the average, shared-nothing architectures beget extreme resource underutilization, high cost and inflexibility. Disaggregating compute from storage has the potential to overcome these limitations! To realize this goal, we are working along several directions:
$3M award from NSF, Google faculty research award, Open-sourced, Deployed in the real world.
- Host Networking for Terabit Ethernet: Rapid adoption of high-bandwidth access links, coupled with relatively stagnant technology trends for essentially all other host resources (core speeds and counts, memory access latencies, cache sizes, etc.), mark a fundamental shift in end-to-end latency and throughput bottlenecks. We are building an understanding of this shift, and designing and building network stacks (within the operating system stack), network protocols, and network hardware for Terabit Ethernet.
Google Research Scholar award.
- Understanding Host Network Stack Overheads [SIGCOMM'21] for Terabit Ethernet.
- Understanding Host Interconnect Congestion [HotNets'22] for Terabit Ethernet.
- NetChannel [SIGCOMM'22], a host network stack architecture for Terabit Ethernet.
- PANCAKE for Secure (Oblivious) Cloud Storage: One of the core problems in offloading data to the cloud is that, even if data is encrypted, an adversary can launch powerful attacks rendering data encryption ineffective. We are designing systems that build upon strong theoretical foundation to enable secure offload of data even under powerful adversarial attacks.
$1M NSF Award, Usenix Security'20 distinguished paper award.
- ShortStack [OSDI'22] for distributed, fault-tolerant, proxy design in oblivious data access
- Pancake [Security'20] for efficient oblivious data access under passive persistent adversaries
- Obladi [OSDI'18] for transactions on ORAM
- Near-optimal Datacenter Design: We are designing and building datacenter transport designs that provide provable worst-case guarantees. Some of the projects include:
SIGCOMM'18 Best Student Paper Award.
- Formal Methods for Network Performance Analysis [NSDI'23];
- dcPIM [SIGCOMM'22], a near-optimal proactive receiver-driven datacenter transport;
- Oblivious reconfigurable (circuit-switched) networks [STOC'22];
- Throughput-optimal Datacenter Network Scheduling [PODC'22];
- CodedBulk [NSDI'21] for Inter-datacenter bulk transfers;
- Sincronia [SIGCOMM'18] for Coflows;
- Universal Packet Scheduling [NSDI'16] for flexible packet scheduling;
Few Past Projects
Here are some of the projects that I have worked on in the past:
- Succinct, a distributed storage system that enables queries directly on compressed data.
Open-sourced and deployed in the real-world.
- Succinct [NSDI'15] for random access, substring search, and even regular expression matches directly on semi-structured data;
- BlowFish [NSDI'16] that enables a smooth performance-storage tradeoff;
- ZipG [SIGMOD'17] for graph queries directly on compressed graphs.
- Anteater and PathDump, systems for datacenter network data plane monitoring and debugging.
Laid the foundation for research on network monitoring and debugging at the data plane. Open-sourced.
- Anteater [SIGCOMM'11], one of the first systems that proposed network debugging at the data plane.
- PathDump [OSDI'16] for end-hosts based monitoring and debugging;
- SwitchPointer [NSDI'18] for enabling in-network visibility;
- Confluo [NSDI'19] for efficient end-host stacks for low-overhead monitoring and debugging.
- Approximate Distance Oracles and Compact Routing, that introduced new data structures, algorithms and techniques for approximate distance queries on graphs.
This project led to the first improvement over several classical decade-old theory results.
|CCC Workshop on Wide-Area Analytics
|SIGMETRICS Tutorials Chair
|NSDI Poster Chair
|HotOS General Chair
- Program Committee