YoVDO

Stanford Seminar - The Quest for Low Storage Latency Changes Everything

Offered By: Stanford University via YouTube

Tags

Computer Science Courses System Architecture Courses NVMe Courses Persistent Memory Courses

Course Description

Overview

Explore the evolution of storage technology and its impact on latency in this Stanford seminar. Delve into the historical progression from early IBM systems to modern NVMe NAND SSDs, examining the ongoing quest for lower storage latency. Investigate media and platform improvements, including next-generation NVM/SCM technologies like 3D Xpoint. Analyze system architecture advancements, such as NVMe and PCI interfaces, and their role in reducing latency. Examine storage software stack optimizations and the potential of persistent memory. Learn about Intel's DIMM technology based on 3D XPoint, the Open NVM Programming Model, and operating system support for persistent memory. Gain insights into new instructions for flushing writes and see example code demonstrating these concepts.

Syllabus

Introduction.
1956: IBM RAMAC 350.
1983: IBM3380.
2016: 10K RPM HDD.
2016 NVMe NAND SSD.
The Continuing Need For Lower Latency.
Lower Storage Latency Requires Media and Platform Improvements.
Addressing Media Latency: Next Gen NVM/SCM.
3D Xpoint Technology Video.
Demonstration of 3D Xpointâ„¢ SSD Prototype.
Need to Address System Architecture To Go Lower.
Addressing Interface Efficiency With NVMe / PCI.
NVMe Delivers Superior Latency Platform HW/5W Average Latency Excluding Media 4KB.
Storage SW Stack Optimizations.
Synchronous Completion for Queue Depth 1?.
Summary: Block Storage Platform Changes.
Persistent Memory Oriented Platform Changes.
INTEL DIMMS Based on 3D XPoint Technology.
Why Persistent Memory?.
Open NVM Programming Model.
Operating System Support for Persistent Memory.
New Instructions For Flushing Writes.
Example Code.


Taught by

Stanford Online

Tags

Related Courses

Accelerating Apache Spark Shuffle for Data Analytics on Cloud with Remote Persistent Memory Pools
Databricks via YouTube
Promise and Pitfalls of Persistent Memory
Strange Loop Conference via YouTube
Crimson - A New Ceph OSD for the Age of Persistent Memory and Fast NVMe Storage
USENIX via YouTube
Write-Optimized Dynamic Hashing for Persistent Memory
USENIX via YouTube
ctFS - Replacing File Indexing with Hardware Memory Translation through Contiguous File Allocation for Persistent Memory
USENIX via YouTube