BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Black Box Far-memory Aware Scalable Data Structures
DTSTART:20230613T103000
DTEND:20230613T123000
DTSTAMP:20260407T120124Z
UID:784b24e88da5b35c40ebd4951a38f32a0a688a2136db965fdd2c4fd6
CATEGORIES:Conferences - Seminars
DESCRIPTION:Yueyang Pan\nEDIC candidacy exam\nExam president: Prof. James 
 Larus\nThesis advisor: Prof. Sanidhya Kashyap\nCo-examiner: Prof. George C
 andea\n\nAbstract\nMemory Disaggregation architecture physically separates
 \nCPU and memory into different components\, usually\nconnected via high-s
 peed network such as RDMA. The concept\nof disaggregated memory was propos
 ed more than a decade\nago. Recently\, it has regained attention because o
 f the memory\nwall\, the imbalanced resource utilization in the data cente
 rs and\nthe emergence of the new hardware. Different approaches have\nbeen
  proposed to utilize disaggregated memory with different\nefficiency and g
 enerality. This report first examines how Fastswap\nutilizes transparently
  disaggregated memory via the kernel\nswapping mechanism. Then it summariz
 es AIFM\, a userspace\nframework for data structure developers that tries 
 to reduce\nthe overhead of the kernel mechanism. Finally it analyzes how\n
 Sherman\, a specialized B+Tree\, achieves performance under\ncontended wor
 kloads. The report ends with the research proposal\nto develop a general d
 ata structure development framework on\ndisaggregated memory.\n\nBackgroun
 d papers\n1. "Can Far Memory Improve Job Throughput?" https://amyousterhou
 t.com/papers/cfm_eurosys20.pdf\n2. AIFM: High-Performance\, Application-In
 tegrated Far Memory https://www.usenix.org/system/files/osdi20-ruan.pdf\n3
 . Sherman: A Write-Optimized Distributed B+Tree Index on Disaggregated Mem
 ory https://dl.acm.org/doi/pdf/10.1145/3514221.3517824.\n\n\n 
LOCATION:BC 229 https://plan.epfl.ch/?room==BC%20229
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR