![]() Princeton University
|
Computer Science 518
|
|
Graduate Coordinator: Melissa Lawson
Discussions: | 20% |
Presentation: | 20% |
Critiques: | 20% |
Project: | 40% |
September 2005
Su Mo Tu We Th Fr Sa
1 2 3
4 5 6
7 8 9 10
11 12 13 14 15 16 17
18 19 20 21 22 23 24
25 26 27 28 29 30
15 - | Organizational meeting |
20 - | Vivek Pai Flash & Harvest |
22 - | pres #1 - Vivek presents DeBox |
27 - | pres #2 - Yi presents CDN |
29 - | (bumped) NSF Grand Challenges Workshop |
04 - | pres #3 - Chris presents ESX |
06 - | pres #4 - Janet presents Lottery |
07 - | pres #5 - Martin presents Resource Makeup lecture |
11 - | pres #6 - Chang presents LARD |
13 - | pres #7 - Wei presents DSM |
18 - | pres #8 - Spencer presents XFS |
20 - | pres #9 - Xiaojuan presents GMS |
25 - |
pres #10 - Stephen presents L4 |
27 - | pres #11 - Janet presents CFS |
November 2005
Su Mo Tu We Th Fr Sa
1 2 3 4 5
6 7 8
9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29 30
01 - |
Fall Recess |
03 - |
Fall Recess |
08 - |
Project proposals |
10 - |
pres #12 - Martin presents UVM |
15 - |
pres #13 - Stephen presents IO-Lite |
17 - |
pres #14 - Yi presents SEDA |
22 - |
pres #15 - Chang presents kqueue |
24 - |
Thanksgiving recess |
29 - |
pres #16 - Chris presents Energy |
December 2005
Su Mo Tu We Th Fr Sa
1 2 3
4 5 6 7 8 9 10
11 12 13 14 15 16 17
18 19 20 21 22 23 24
25 26 27 28 29 30 31
01 - |
pres #17 - Xiaojuan presents Xen |
02 - |
pres #18 - Will presents Spin Makeup lecture |
06 - |
pres #19 - Wei presents Metal |
08 - |
pres #20 - Spencer presents soft-updates |
09 - |
pres #21 - Will presents Byzantine Makeup lecture |
13 - |
(bumped) WORLDS workshop |
15 - |
(bumped) FAST conference |
10 - | Project presentations |
12 - |
Project presentations |
17 - | Project presentations Dean's Date |
Byzantine & RAID - making systems fault tolerant
CFS & NFS - distributed filesystems, LAN and WAN
CDN & Chash - content distribution networks and consistent hashing
DSM & RPC - distributed shared memory and remote procedure calls
Debox & DTrace - kernel debugging
Energy & DVS - energy efficiency and dynamic voltage scaling
ESX & DISCO - virtual machines
Flash & Harvest - Fast servers - web and proxy
IO-Lite & fbufs - buffering systems inside the OS
kqueue & select - fast event notification systems
L4 & Exokernel - stripped down kernels
LARD & Network Dispatcher - cluster request distribution
Metal & Synthesis - error detection, dynamic code generation
resource & scout - kernel resource management
SEDA & Scheduler Activations - managing parallelism
Soft-Updates & LFS - file system restructuring
SPIN & VINO - extensible kernels
UVM & Mach - virtual memory restructuring
Xen & Denali - virtualization, paravirtualization
XFS & FFS - advanced file systems
Worldwide timing - clock skew becomes an issue when running distributed applications. Design a system to provide a more coherent global clock.
Scalability - test some aspect of OS scalability - threads, zero copy, virtual memory design, etc. Fix it if possible.
Interface - the select system call does not like disk file descriptors. Make it useful by having it begin prefetching when presented with a disk fd.
Anomaly detection - by observing the behaviors of large numbers of somewhat similar programs, can you determine when a program has gone out of control?
geolocation - can you tell where someone is coming from, and redirect them to the closest instance of a service?
Page faults - can you lay out code or data to reduce fault rates? Does it matter?
Virtualization costs - can you reduce the overhead of virtualization by making changes at the program level?
Do class-based resource management - see Marc Fiuczynski's project 3 for a more complete description.