CSE590S
Fall 2003
Overview: In ten years time, the storage available on conventional computer systems and digital appliances will reach hundreds of terabytes, if not petabytes Even today, mass storage is selling for roughly 1 dollar/Gigabyte, making it stunningly inexpensive to equip a machine with a huge amount of storage. In response to this, weÕre seeing new developments in the continuing process of digital convergence, which include Personal Video Recorders such as TIVO, and legitimate music and video services. At the same time, bandwidth providers are working to deliver ever increased bandwidth to the home and business through ISDN, cable, fixed wireless, and even satellite. Together, these two trends (big disks, fat pipes) promise to dramatically change the way in which digital content (eg, TV) is stored, delivered, and managed at the end node, within the network, and at the source.
Objective: In this quarterÕs 590s, we will examine STORAGE and PIPE technology trends in order to understand where todayÕs system architectures will break down, and the new problems that will need to be solved in the next ten years. In addition, we will also look at materials outside of core computer science in order to gain a better appreciation for the generality of such phenomena as Ònetwork effectsÓ and Òzipfiness.Ó Lastly, weÕll take a look at some early entrants into the ÒBIG STORAGE+BIG PIPE==BIG FUNÓ space.
Operations: WeÕre going to run things a little differently this quarter. Rather than assign an individual to read and present each weekÕs paper (often making them one of the few who have read the paper), weÕre going to organize much more like a moderated discussion. Each week weÕll have a moderator who will prepare a set of ÒprovocativeÓ questions/issues surround that weekÕs material. The actual content of the discussion comes from the class participants, which is presumably based on their thoughtful reading of the materials. The moderatorÕs job will be to move things along when they stall, and to get things back on track when they go off track.
As with previous quarters, we expect that you will have read the first paper by the second session. In the first section, weÕll discuss the reading list and present a little bit of background material to help motivate why weÕre looking at this topic.
The reading list is broken into three major sections with the principal papers for each week shown. In addition, background materials are also offered. Like wheaties and spinach, you should read these since it would be good for you.
Proceedings of the 100th
Anniversary Conference of Magnetic Recording and Information Storage. In particular, you might want to focus
on the talk by John Best (ÒThe Current 60% CGR in Desnity and the FutureÓ)
The
1998 WTEC Workshop on the Future of Data Storage Technologies has its slides
online:
http://www.wtec.org/loyola/hdmem/views/top.htm
FAST
2002 keynotes had technical content on disk technology: http://www.usenix.org/events/fast02/.
In particular,
you should check out the Morris keynote: ÒFrom Atoms to PeopleÓ.
October
10. The near term future of storage technologies.
Technical Impact of Magnetic Hard Disk Drives on Storage
Systems. Ed Growchowski. 2003.
Emerging Trends in Storage on Magnetic Hard Disk
Drives. Edward Grochowski.
1998.
A Conversation with Jim Gray. CACM Queue. June
2003.
TeraScale SneakerNet: Using Inexpensive Disks
for Backup, Archiving, and Data Exchange. Jim Gray, Wyman Chong, Tom Barclay, Alex Szalay, Jan
Vandenberg. May 2002. MS TR 02 54.
NOTE: We need a paper that discusses large scale filing on a single system.
NOTE: SOSP is this week. Depending on how many people are here, we may or may not meet.
Note: We will be discussing A BOOK. You should start reading early starts to ensure that you have time to complete it.
How Much Information is there in the World. Michael Lesk.
Internet TV: Implications for the long distance network. AM Odlyzko. 2001.
Other Relevant Papers. Available on the web.
Information Rules: A Strategic Guide to the Network Economy. Available at Amazon. Carl Shapiro and Hal Varian. November 1998.
October 24. The Real World
How Much Information Ð 2000. Peter Lyman and Hal Varian. Also available in html here. Warning: This is long and dense, but filled with great information.
Gordon MooreÕs Personal Cyberstore.
A Personal Store for Everything (PDF) . Gordon Moore. Published in the January 2001 issue of the Communications of the ACM.
October 31.
The Internet World
Measurement, Modeling and Analysis of a Peer to Peer File Sharing Workload. Gummada, Dunn, Saroiu, Gribble, Levy, Zahorjan. SOSP 2003.
Growth of the Internet. Coffman and Odlyzko. 2001.
The History of Communications and its Implications for the Internet. Odlyzko. 2000.
Content is Not King. Odlyzko. 2001.
November 7.
Al GoreÕs Internet did not create ZipfÕs Law
Linked: The New Science of Networks. (Available at Amazon). Albert-LasÕloÕ BarabaÕsi.
Background
Broadband Alphabet Soup Tutorial
The Role of Satellites. A BureaucratÕs EyeÕ
Filling
the Communications Gap.
RFC 3077. A Link Level Tunneling
Mechanism for Unidirectional Links.
IP Based enhanced Data Casting
Services over Radio Networks.
Kellerer, Spiers, Eberspacher. 2001.
TCP Over Assymetric
Networks (Slides)
Jannotti et al, "Overcast: Reliable
Multicasting with an Overlay Network"
Chu, Rao, and Zhang: A Case for End System Multicast" [There is a JSAC and a SIGMETRICS paper. the JSAC is better, but longer]
November 28. HOLIDAY.
December 5. Media Appliances
Tivo Architecture Overview (Slides)