Systems Software in the New Computing World

Speaker: Dr. Diyu ZHOU
         EPFL

Title:  "Systems Software in the New Computing World"

Date:   Thursday, 9 March 2023

Time:   3:00pm - 4:00pm HKT

Zoom link:
https://hkust.zoom.us/j/465698645?pwd=aVRaNWs2RHNFcXpnWGlkR05wTTk3UT09

Meeting ID: 465 698 645
Passcode: 20222023


Abstract:

Exponential growth in users, requests, and data poses an ever increasing
demand on the performance of today's data centers. This challenge has
resulted in two major trends. First, data centers scale out the
computation by leveraging multicore architecture and deploying more
servers. Second, ultra-fast storage devices are developed to meet the
exponential growth in data. Unfortunately, traditional systems software is
a poor fit for these trends, rendering applications unable to realize the
potential of these developments.

In this talk, I will present my work on designing modern system software
to exploit these computing trends by supporting three critical application
requirements: I/O efficiency, multicore scalability, and practical
reliability. I will first present OdinFS, a high-performance and scalable
file system for emerging non-volatile memory (NVM). By taking into account
the unique characteristics of NVM, OdinFS scales to hundreds of cores and
achieves tens to hundreds of times better performance than prior state of
the art. I will next present RRC, an application-transparent replication
system for commercial off-the-shelf containers. RRC incurs latency
overhead up to 75x lower than competitive schemes, while also achieving
significantly lower throughput overhead, thus enabling practical
deployment for critical server applications.


*******************
Biography:

Diyu Zhou is postdoctoral researcher at EPFL. He completed his Ph.D. at
UCLA advised by Yuval Tamir. His research focuses on building
high-performance, scalable, and reliable computer systems. Specifically,
he has developed I/O stacks to support modern storage devices, devised
frameworks and algorithms for synchronization primitives to scale to
massive multi-core machines, found and fixed concurrency bugs, and
designed practical fault tolerance mechanisms for modern systems.