USENIX Courses

Light-Dedup - A Light-weight Inline Deduplication Framework for Non-Volatile Memory File Systems

Explore a lightweight deduplication framework for NVM file systems, featuring innovative techniques to enhance performance and reduce costs in next-generation storage media.

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

SingularFS - A Billion-Scale Distributed File System Using a Single Metadata Server

Innovative distributed file system using a single metadata server for billion-scale operations, featuring log-free metadata, hierarchical concurrency control, and hybrid inode partition.

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

Prefix Siphoning - Exploiting LSM-Tree Range Filters for Information Disclosure

Explores key disclosure timing attacks on LSM-tree based key-value stores using range filters, demonstrating security vulnerabilities in data storage systems.

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

STYX - Exploiting SmartNIC Capability to Reduce Datacenter Memory Tax

Innovative framework offloading memory optimization tasks to SmartNICs, reducing server CPU load and cache pollution while maintaining efficiency. Demonstrates significant latency improvements for co-running applications.

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

Cyclosa - Redundancy-Free Graph Pattern Mining via Set Dataflow

Innovative framework for efficient graph pattern mining, eliminating redundancies through set dataflow representation and parallel execution, outperforming existing systems in various scenarios.

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

Oakestra - A Lightweight Hierarchical Orchestration Framework for Edge Computing

Lightweight, hierarchical orchestration framework for edge computing, offering reduced resource usage and improved performance through federated management and semantic overlay networking.

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

Adaptive Online Cache Capacity Optimization via Lightweight Working Set Size Estimation at Scale

Explore Cuki, an innovative data structure for efficient cache optimization in big data applications, improving accuracy and performance in distributed file caching systems.

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

Nodens - Enabling Resource Efficient and Fast QoS Recovery of Dynamic Microservice Applications in Datacenters

Explore Nodens, a runtime system for fast QoS recovery in dynamic microservice applications, featuring traffic-based monitoring, blocking-aware load updating, and efficient query draining.

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

Towards Iterative Relational Algebra on the GPU

Explores GPU-based iterative relational algebra for high-performance data analytics, presenting a novel hash-join implementation with optimizations for memory access and deduplication, outperforming existing solutions.

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

Deploying User-space TCP at Cloud Scale with LUNA

Explore LUNA, a high-performance user-space TCP stack deployed at Alibaba Cloud, offering significant improvements in throughput and latency compared to kernel-based and other user-space solutions.

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

TC-GNN - Bridging Sparse GNN Computation and Dense Tensor Cores on GPUs

Innovative framework accelerating graph neural networks using GPU Tensor Core Units, reconciling sparse GNN computation with dense TCUs for improved performance in graph-based machine learning.

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

Revisiting Secondary Indexing in LSM-based Storage Systems with Persistent Memory

Explores PERSEID, an efficient secondary indexing mechanism for LSM-based storage systems using persistent memory, offering improved performance over existing techniques.

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

Avoiding the Ordering Trap in Systems Performance Measurement

Explores the impact of trial ordering on systems performance experiments, revealing hidden biases and proposing methods to detect and mitigate ordering effects in research studies.

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

Luci - Loader-based Dynamic Software Updates for Off-the-shelf Shared Objects

Explore Luci, a lightweight technique for dynamically updating shared libraries during runtime without service interruptions or source modifications, enhancing security and availability in software systems.

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

MELF - Multivariant Executables for a Heterogeneous World

Explore MELF: a novel approach for including multiple function variants in a single binary, enabling runtime switching for optimized performance across heterogeneous systems.

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