USENIX Courses

Why Are Distributed Systems So Hard?

Explore the challenges of distributed systems, including CAP theorem myths, network partitions, consensus algorithms, and human factors, with cat-illustrated examples.

• YouTube
• 28 minutes
• Self-Paced
• Conference Talk

Tracing, Fast and Slow - Digging into and Improving Your Web Service's Performance

Explore techniques for tracing and improving web service performance, including black & white box tracing, distributed systems tracing, and various tools to measure and enhance service efficiency.

• YouTube
• 30 minutes
• Self-Paced
• Conference Talk

Shipping Software with an SRE Mindset

Explore how SRE techniques can be applied to on-premise software delivery, enhancing construction, packaging, and shipment processes for improved reliability and customer satisfaction.

• YouTube
• 27 minutes
• Self-Paced
• Conference Talk

Lightning Talks

Lightning talks on diverse SRE topics: incident response, distributed systems, resource management, automation, operational maturity, and monitoring. Insights from industry experts on improving reliability and efficiency.

• YouTube
• 1 hour 9 minutes
• Self-Paced
• Conference Talk

Fault Tree Analysis Applied to Apache Kafka

Explore fault tree analysis techniques to enhance Apache Kafka cluster resilience, drawing insights from Lyft's experience in bulletproofing their deployments.

• YouTube
• 31 minutes
• Self-Paced
• Conference Talk

Latency SLOs Done Right

Explore correct approaches to calculating latency SLOs using histograms and mathematically accurate quantiles, improving service level objectives for more effective monitoring and performance management.

• YouTube
• 31 minutes
• Self-Paced
• Conference Talk

OpenEC - Toward Unified and Configurable Erasure Coding Management in Distributed Storage Systems

Unified framework for deploying erasure coding solutions in distributed storage systems, simplifying integration and optimizing repair performance.

• YouTube
• 24 minutes
• Self-Paced
• Conference Talk

SPEICHER - Securing LSM-based Key-Value Stores Using Shielded Execution

Secure key-value storage system using shielded execution, providing confidentiality, integrity, and data freshness. Implements authenticated LSM data structure and asynchronous trusted counters for enhanced security on untrusted hosts.

• YouTube
• 21 minutes
• Self-Paced
• Conference Talk

Orion - A Distributed File System for Non-Volatile Main Memory and RDMA-Capable Networks

Explores a novel distributed file system design optimized for non-volatile main memory and RDMA networks, achieving high performance for metadata and data access while maintaining byte addressability.

• YouTube
• 30 minutes
• Self-Paced
• Conference Talk

FlowBlaze - Stateful Packet Processing in Hardware

Innovative hardware-based solution for stateful packet processing, enabling high-speed network functions with low latency and power consumption using Extended Finite State Machines on programmable NICs.

• YouTube
• 26 minutes
• Self-Paced
• Conference Talk

Blink - Fast Connectivity Recovery Entirely in the Data Plane

Explores programmable switches for rapid rerouting using TCP-induced signals to detect failures. Presents Blink, a data-driven system analyzing TCP flows at line rate to quickly recover connectivity in the data plane.

• YouTube
• 24 minutes
• Self-Paced
• Conference Talk

Size-Aware Sharding for Improving Tail Latencies in In-Memory Key-Value Stores

Innovative size-aware sharding technique for in-memory key-value stores, reducing tail latencies and improving throughput by avoiding head-of-line blocking and optimizing request distribution across cores.

• YouTube
• 23 minutes
• Self-Paced
• Conference Talk

BLAS-on-Flash - An Efficient Alternative for Large Scale ML Training and Inference

Efficient alternative for large-scale ML tasks using SSD-resident matrices, enabling near in-memory performance on a single workstation. Outperforms distributed systems for complex algorithms like eigensolvers.

• YouTube
• 24 minutes
• Self-Paced
• Conference Talk

Chaos Engineering

Explore Netflix's approach to chaos engineering, its impact on system resilience, and lessons learned in building a robust software stack through controlled failure experiments.

• YouTube
• 38 minutes
• Self-Paced
• Conference Talk

Debugging Linux Issues with eBPF

Explore eBPF for debugging Linux performance issues, focusing on a real-world case of CPU increase in Kafka after a Debian upgrade. Learn tools and techniques for operational engineers.

• YouTube
• 27 minutes
• Self-Paced
• Conference Talk