AI Adoption - Drive Business Value and Organizational Impact
Free courses from frontend to fullstack and AI
Overview
Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
This conference talk presents AegonKV, a novel key-value separated LSM store that simultaneously addresses three critical challenges in storage systems: throughput, tail latency, and space usage. Learn how researchers from Huazhong University of Science and Technology developed a "three-birds-one-stone" solution that leverages SmartSSD-based garbage collection offloading to enhance performance without competing for CPU or I/O resources. The presentation demonstrates how AegonKV achieves 1.28-3.3 times throughput improvement, 37%-66% reduction in tail latency, and 15%-85% reduction in space overhead compared to existing KV separated systems. Discover the offload-friendly data structures and hardware/software execution logic that overcome the challenges of garbage collection in key-value separated storage systems.
Syllabus
FAST '25 - AegonKV: A High Bandwidth, Low Tail Latency, and Low Storage Cost KV-Separated LSM Store
Taught by
USENIX