USENIX Courses

SwiftPaxos - Fast Geo-Replicated State Machines

Explore SwiftPaxos, a protocol enhancing geo-replicated state machines with improved latency and throughput, outperforming Paxos in high-contention scenarios through innovative voting mechanisms.

• YouTube
• 16 minutes
• Self-Paced
• Free Video

NetAssistant: Dialogue-Based Network Diagnosis in Data Center Networks

Dialogue-based AI system for efficient network diagnosis in data centers, reducing manual oncalls and improving issue resolution through natural language interactions and automated workflows.

• YouTube
• 18 minutes
• Self-Paced
• Free Video

Netcastle: Network Infrastructure Testing at Scale

Netcastle: A comprehensive CI/CD network testing system addressing reliability challenges in complex infrastructures. Supports various use-cases across multiple teams, running 500K+ daily tests for firmware, control planes, and controllers.

• YouTube
• 19 minutes
• Self-Paced
• Free Video

VILAM: Infrastructure-assisted 3D Visual Localization and Mapping for Autonomous Driving

Innovative framework leveraging roadside infrastructure for high-precision localization and mapping in autonomous vehicles, achieving decimeter-level accuracy with consumer-grade cameras across diverse scenarios.

• YouTube
• 15 minutes
• Self-Paced
• Free Video

Horus - Granular In-Network Task Scheduler for Cloud Datacenters

Innovative in-network task scheduler for cloud datacenters, optimizing short-lived task distribution across workers to minimize response time and maximize throughput at scale.

• YouTube
• 18 minutes
• Self-Paced
• Free Video

Revisiting Congestion Control for Lossless Ethernet

Explores ACK-driven congestion control for lossless Ethernet, improving flow completion time and addressing issues like congestion spreading and head-of-line blocking through packet conservation principles.

• YouTube
• 18 minutes
• Self-Paced
• Free Video

Swing - Short-cutting Rings for Higher Bandwidth Allreduce

Innovative algorithm enhances allreduce performance on torus networks, reducing communication hops and improving bandwidth by up to 3x for various data sizes and network configurations.

• YouTube
• 19 minutes
• Self-Paced
• Free Video

POSEIDON: A Consolidated Virtual Network Controller for Managing Millions of Tenants

Innovative virtual network control framework enhancing cloud performance through consolidated functions, tree-based abstraction, and efficient config management, resulting in significant throughput increase and latency reduction.

• YouTube
• 17 minutes
• Self-Paced
• Free Video

MESSI: Behavioral Testing of BGP Implementations

Automated testing tool for BGP implementations, generating 150K tests to identify correctness bugs in popular routing software. Addresses challenges in stateful protocols and complex structures.

• YouTube
• 18 minutes
• Self-Paced
• Free Video

AutoSketch: Automatic Sketch-Oriented Compiler for Query-driven Network Telemetry

Innovative compiler integrating query-driven telemetry and sketch algorithms for efficient network monitoring, balancing expressiveness, accuracy, and resource usage through automated optimization.

• YouTube
• 15 minutes
• Self-Paced
• Free Video

TANGO: Secure Collaborative Route Control across the Public Internet

Secure system enabling smaller networks to optimize Internet routing without relying on large providers. Exposes diverse paths, collects telemetry, and dynamically reroutes traffic to improve performance.

• YouTube
• 14 minutes
• Self-Paced
• Free Video

Flow Scheduling with Imprecise Knowledge in Data Center Networks

Innovative flow scheduling solution QCLIMB minimizes completion times in data center networks using imprecise flow information, outperforming existing methods without switch modifications.

• YouTube
• 16 minutes
• Self-Paced
• Free Video

Leo: Online ML-based Traffic Classification at Multi-Terabit Line Rate

Innovative system for high-speed online traffic classification using ML in network switches. Enables real-time analysis for security, QoS, and IoT applications at multi-terabit rates with high accuracy and flexibility.

• YouTube
• 18 minutes
• Self-Paced
• Free Video

LoLKV - Logless, Linearizable, RDMA-based Key-Value Storage System

Innovative key-value storage system using lock-free design for concurrent updates, achieving higher throughput and lower latency than state-of-the-art alternatives.

• YouTube
• 14 minutes
• Self-Paced
• Free Video

Making Kernel Bypass Practical for the Cloud with Junction

Junction: A kernel bypass system enabling dense packing of cloud instances while maintaining Linux app compatibility. Improves throughput and latency without code changes, supporting various runtimes and outperforming native Linux.

• YouTube
• 16 minutes
• Self-Paced
• Free Video