Effectively Prefetching Remote Memory with Leap
Offered By: USENIX via YouTube
Course Description
Overview
Explore a conference talk from USENIX ATC '20 that introduces Leap, an innovative prefetching solution for remote memory accesses in memory disaggregation systems. Learn how Leap employs an online, majority-based prefetching algorithm to increase page cache hit rates and implements a lightweight, efficient data path in the kernel to mitigate latency bottlenecks. Discover how this approach significantly improves median and tail remote page access latencies for memory-bound applications, leading to substantial performance enhancements compared to state-of-the-art solutions. Gain insights into the challenges of memory disaggregation over RDMA and how Leap addresses them to optimize application performance in memory-constrained environments.
Syllabus
USENIX ATC '20 - Effectively Prefetching Remote Memory with Leap
Taught by
USENIX
Related Courses
Windows Server 2019: Advanced Networking FeaturesLinkedIn Learning Deep Dive into GPU Support in Apache Spark 3.x - Accelerator-Aware Scheduling and RAPIDS Plugin
Databricks via YouTube Microsecond Consensus for Microsecond Applications
USENIX via YouTube An Edge-Queued Datagram Service for All Datacenter Traffic
USENIX via YouTube Building a High Performance Network in the Public Cloud Using RDMA - First Principles
Oracle via YouTube