Publications

MTP: Transport for In-Network Computing

Portable and High-Performance SmartNIC Programs with Alkali

Expressive and Efficient Service Mesh Policies

Read-ME: Refactorizing LLMs as Router-Decoupled Mixture of Experts with System Co-Design

MOSEL: Inference Serving Using Dynamic Modality Selection

FFN-SkipLLM: A Hidden Gem for Autoregressive Decoding with Adaptive Feed Forward Skipping

Optimizing Transformer Inference with Selective Distillation: Layerwise Conversion to Linear Attention

On the Criticality of Integrity Protection in 5G Fronthaul Networks

ChainedFilter: Combining Membership Filters by Chain Rule

Cassini: Network-Aware Job Scheduling in Machine Learning Clusters

On a Foundation Model for Operating Systems

Yama: Providing Performance Isolation for Black-Box Offloads

Enabling Resilience in Virtualized RANs with Atlas

LogNIC: A High-Level Performance Model for SmartNICs

Configuring the OS for Tomorrow's Robots

Resilient Baseband Processing in Virtualized RANs with Slingshot

Darwin: Flexible Learning-based CDN Caching

Blink-hash: An Adaptive Hybrid Index for In-Memory Time-Series Databases

Navigating Performance-Efficiency Tradeoffs in Serverless Computing: Deduplication to the Rescue!

Lowering the Pre-training Tax for Gradient-based Subset Training: A Lightweight Distributed Pre-Training Toolkit

Invited Paper: Towards Efficient Microservice Communication

Better Together: Jointly Optimizing ML Collective Scheduling and Execution Planning using SYNDICATE

Shockwave: Fair and Efficient Cluster Scheduling for Dynamic Adaptation in Machine Learning

RingLeader: Efficiently Offloading Intra-Server Orchestration to NICs

Sketchovsky: Enabling Ensembles of Sketches on Programmable Switches

ExoPlane: An Operating System for On-Rack Switch Resource Augmentation

Towards a Machine Learning-Assisted Kernel with LAKE

Towards Accelerating Data Intensive Application's Shuffle Process Using SmartNICs

Congestion Control in Machine Learning Clusters

Multi-agent Databases via Independent Learning

Impact of RoCE Congestion Control Policies on Distributed Training of DNNs

Think before you shuffle: data-driven shuffles for geo-distributed analytics

Jiffy: Elastic Far-Memory for Stateful Serverless Analytics

Memory Deduplication for Serverless Computing with Medes

Elastic Model Aggregation with Parameter Service

Doing more by doing less: how structured partial backpropagation improves deep learning clusters

PL2: Towards Predictable Low Latency in Rack-Scale Networks

TCP is Harmful to In-Network Computing: Designing a Message-Oriented Transport Protocol (MTP)

A Vision for Runtime Programmable Networks

Atoll: A Scalable Low-Latency Serverless Platform

D2R: Policy-Compliant Fast Reroute

CliqueMap: Productionizing an RMA-Based Distributed Caching System

Running BGP in Data Centers at Scale

ATP: In-network Aggregation for Multi-tenant Learning

Whiz: Data-Driven Analytics Execution

Accelerating Deep Learning Inference via Learned Caches

AED: Incrementally Synthesizing Policy-Compliant and Manageable Configurations

PANIC: A High-Performance Programmable NIC for Multi-tenant Networks

SNF: Serverless Network Functions

Network-accelerated Distributed Machine Learning for Multi-Tenant Settings

1RMA: Re-envisioning Remote Memory Access for Multi-tenant Datacenters

Detecting Network Load Violations for Distributed Control Planes

Tiramisu: Fast and General Network Verification

Themis: Fair and Efficient GPU Cluster Scheduling for Machine Learning Workloads

Liveness Verification of Stateful Networks.

Automated Verification of Customizable Middlebox Properties with Gravel

Smurf: Self-Service String Matching Using Random Forests

On the Impact of Cluster Configuration on RoCE Application Design

Accelerating Deep Learning Inference via Freezing

The Design and Operation of a Platform for Advanced Cloud Experimentation

Loom: Flexible and Efficient NIC Packet Scheduling

Correctness and Performance for Stateful Chained Network Functions

Synthesis of Fault-Tolerant Distributed Router Configurations

Your Programmable NIC Should be a Programmable Switch

Dynamic Query Re-Planning using QOOP

Iron: Isolating Network-based CPU in Container Environments

Monarch: Gaining Command on Geo-Distributed Graph Analytics

Bridging the GAP: Towards Approximate Graph Analytics

RoGUE: RDMA over Generic Unconverged Ethernet

Low Latency Software Rate Limiters for Cloud Networks

P5: Policy-driven optimization of P4 pipeline

Genesis: Synthesizing Forwarding Tables in Multi-tenant Networks

UNO: Unifying Host and Smart NIC Offload for Flexible Packet Processing

Automatically Repairing Network Control Planes Using an Abstract Representation

Granular Computing and Network Intensive Applications: Friends or Foes?

Supporting Diverse Dynamic Intent-based Policies using Janus

Bootstrapping Evolvability for Inter-domain Routing with D-BGP

Titan: Fair Packet Scheduling for Commodity Multiqueue NICs

Fast Control Plane Analysis Using an Abstract Representation

AC/DC TCP: Virtual Switch-based Congestion Control Enforcement for Datacenter Networks

Paving the Way for NFV: Simplifying Middlebox Modifications using StateAlyzr

A Standardized Southbound API for VNF Management

Packing and Dependency-aware Scheduling for Data-Parallel Clusters

Efficiently delivering online services over integrated infrastructure

Clarinet: WAN-Aware Optimization for Analytics Queries

Altruistic Scheduling in Multi-Resource Clusters

A Universal Approach to Data Center Network Design

Latency in Software Defined Networks: Measurements and Mitigation Techniques

extended abstract

Presto: Edge-based Load Balancing for Fast Datacenter Networks

PGA: Using graphs to express and automatically reconcile network policies

PerfSight: Performance Diagnosis for Software Dataplanes

Improving the Safety, Scalability, and Efficiency of Network Function State Transfers

Management Plane Analytics

Improving the performance of SPDY for mobile devices

Measuring Control Plane Latency in SDN-enabled Switches

Low Latency Geo-distributed Data Analytics

Experimenting with Next-Generation Cloud Architectures Using CloudLab

Seeing Through Network Protocol Obfuscation

Bootstrapping Evolvability for Inter-Domain Routing

XIA: Architecting a More Trustworthy and Evolvable Internet

WhoWas: A Platform for Measuring Web Deployments on IaaS Clouds

Multi-Resource Packing for Cluster Schedulers

Design Patterns for Tunable and Efficient SSD based Indexes

OpenNF: Enabling Innovation in Network Function Control

A Call to Arms for Management Plane Analytics

A Highly Available Software Defined Fabric

Developing a Predictive Model of Quality of Experience for Internet Video

Virtual Network Diagnosis as a Service

Understanding Internet Video Viewing Behavior In the Wild

extended abstract

Analyzing the Potential Benefits of CDN Augmentation Strategies for Internet Video Workloads

An Information-Aware QoE-Centric Mobile Video Cache

Enhancing multimedia accessibility and availability with information-bound references

FCP: A Flexible Transport Framework for Accommodating Diversity

Next Stop, the Cloud: Understanding Modern Web Service Deployment in EC2 and Azure

Adaptive Data Transmission in the Cloud

XIA: Efficient Support for Evolvable Internetworking

A Quest for an Internet Video Quality-of-Experience Metric

RPT: Re-architecting Loss Protection for Content-Aware Networks

Toward Software-Defined Middlebox Networking

Challenges in Unifying Control of Middlebox Traversals and Functionality

ECOS: Leveraging Software-Defined Networks to Support Mobile Application Offloading

ECOS: Practical Mobile Application Offloading for Enterprises

DECOR: a Distributed Coordinated Resource Monitoring System

Better Internet Routing Through Intrinsic Support for Selfishness

Obtaining Representative Measurements of Cellular Network Performance

XIA: An Architecture for an Evolvable and Trustworthy Internet

The Evolution of Network Configuration: A Tale of Two Campuses

De-Ossifying Internet Routing through Intrinsic Support for End-Network and ISP Selfishness

Refactoring Content Overhearing to Improve Wireless Performance

MicroTE: Fine Grained Traffic Engineering for Data Centers

Demystifying Configuration Challenges and Trade-offs in Network-based ISP Services

A Comparative Study of Handheld and Non-Handheld Traffic in Campus WiFi Networks

CloudNaaS: A Cloud Networking Platform for Enterprise Applications

Understanding Data Center Traffic Characteristics

Using Strongly Typed Networking to Architect for Tussle

On the Effectiveness of IP reputation for Spam Filtering

Extensible and Scalable Network Monitoring Using OpenSAFE

NashWiFi: Coordinated Management of Heterogenous Home 802.11 Networks Using Bargaining

A Case for Information-Bound Referencing

A First Look at Problems in the Cloud

Tracking Semantic Relationships for Effective Data Management in Home Networks

The Case for Fine-grained Traffic Engineering in Data-centers

EndRE: An End-System Redundancy Elimination Service for Enterprises

Network Traffic Characteristics of Data Centers in the Wild

Peeking into the Cloud: Toward User-driven Cloud Management

Understanding Large-Scale Spamming Botnets From Internet Edge Sites

Cheap and Large CAMs for High-performance Data-intensive Networked Systems

Understanding Data Center Traffic Characteristics

SmartRE: An Architecture for Coordinated Network-wide Redundancy Elimination

Also appeared as a poster at NSDI 2009, Boston, MA

On the Treeness of Internet Latency and Bandwidth

Redundancy in Network Traffic: Findings and Implications

Mining Policies from Enterprise Network Configuration

Phoenix: A System for Automatically Reconfiguring Networks

An Empirical Study of Malware Evolution

Net-Replay: A New Network Primitive

Unraveling the Complexity of Network Management

Router-Level Spam Filtering Using TCP Fingerprints: Architecture and Measurement-Based Evaluation

Packet Caches on Routers: The Implications of Universal Redundant Traffic Elimination

Avoiding File System Micromanagement with Range Writes

Coordinated Configuration of Uncoordinated Wireless Networks: A Win-win Approach Based on Bargaining

On the Performance Benefits of Multihoming Route Control

Remote Profiling of Resource Constraints in Web Servers Using Mini-Flash Crowds

Traffic-Aware Channel Assignment in Enterprise Wireless LANs

Understanding and Exploiting Network Traffic Redundancy

Cooperative Interdomain Traffic Engineering using Nash Bargaining and Decomposition

Using Mini-Flash Crowds to Infer Resource Constraints in Remote Web Servers

A la carte: An Economic Framework for Multi-ISP Service Quality

Cookies Along Trust-Boundaries (CAT): Accurate and Deployable Flood Protection

SANE: A Protection Architecture for Enterprise Networks

Flow Cookies: Using Bandwidth Amplification to Defend Against DDoS Flooding Attacks

short paper

Self-Management in Chaotic Wireless Deployments: Extended Version

Invited paper: Wireless Networks (WINET) Journal, Special Issue on Selected Papers from MobiCom 2005.

Achieving Good End-to-end Service Using Bill-Pay

Endpoint-Based Routing Strategies for Improving Internet Peformance and Resilience

Self Management in Chaotic Wireless Deployments

On the Scaling of Congestion in the Internet Graph

Multihoming Performance Benefits: An Experimental Evaluation of Practical Enterprise Strategies

Availability, Usage, and Deployment Characteristics of the Domain Name System

A Comparison of Overlay Routing and Multihoming Route Control

On the Responsiveness of DNS-based Network Control

Understanding the Impact of Route Control Product Deployment

Invited paper

An Empirical Evaluation of Wide-Area Internet Bottlenecks

Also appeared as a short paper in SIGMETRICS 2003, San Diego, CA

Towards Representative Internet Measurements

Scaling Properties of the Internet Graph

A Measurement-Based Analysis of Multihoming

Detecting DDoS Attacks on ISP Networks

The Impact of False Sharing on Shared Congestion Management

Also appeared in Proc. SIGCOMM Computer Communication Review, January 2002, as a short paper

Selfish Behavior and Stability of the Internet: A Game-Theoretic Analysis of TCP

Multi-Modal Network Protocols: Adapting to Highly Variable Operating Conditions

Also appeared in Proc. SIGCOMM Computer Communication Review, January 2002, as a short paper

Mechanisms for Internet Routing: A Study

Exploring Congestion Control

The Impact of False Sharing on Shared Congestion Management