I work for IBM Corporation's
Almaden Research Center from
Vashon, WA. For contacting me email is best.
I am working on file systems, lately improving filesets and snapshots in
GPFS. I've been working on GPFS
since 2005. Earlier I worked on the
Distributed
Storage Tank project and on Global Namespaces. I was a co-author on
a paper in the IBM
Systems Journal titled
Global namespace for
files
. I am also interested in namespace work in
Grid and
NFSv4
protocols.
Previously I worked at the IBM Pittsburgh Lab, formerly known as Transarc Corporation. Much of the time at Transarc I was working on Episode, the DFS Server's file system. The Winter 1992 Usenix conference contained a fairly comprehensive description of Episode. I am also a co-author of an early paper on DFS, then called DEcorum. Here are some press releases announcing the formation of Transarc Corporation.
Before that I worked on AFS®, specifically the kaserver which implements the Kerberos 4 authentication protocol and an encyption algorithm called FCrypt. The goal of the FCrypt design was to provide a faster, smaller alternative to DES for use in the kernel. In retrospect this, was a mistake; amateur cryptographers should not design new ciphers for production systems. It is much harder to do a good job than it first appears. AFS is now available as open source and is supported by a vigorous development community. I occasionally contribute to the AFSLore Wiki, and I've become a pretty big fan of Wikis generally.
Before moving to Pittsburgh to work on AFS, I worked at Lawrence Livermore National Laboratory on the S-1 project. Mark Smotherman has put together an excellent history of the S-1 Supercomputer. Mark's page had many excellent references to documents from this project. Here are some others from that era:
Adding content to the web is an important contribution that everyone can make to universal knowledge. More specifically, I want to return the favor that Google does me by being such a great search engine. In large part it does this using PageRank, which considers how frequently a page is a link target. But how will this work unless people create links to pages they like? Maintaining a weblog seems to be beyond me, so perhaps a simpler and more likely to succeed approach is a tagged and annotated bookmark system. I am using del.icio.us and hope to add a bookmark there with some tags and a sentence or two about every web page I like that I take the time to read. Here's a copy that will allow Google to index it (see del.icio.us's robots.txt; thanks to Boaz Shmueli for pointing this out). Also, here is my weblog, though I don't update it anymore.
It was brought to my attention, however, that the robots exclusion at del.icio.us is such that the bookmarks there do not contribute to the PageRank of the targets. Thus, my copy of my del.icio.us bookmarks will have to suffice for now.
My vision is to create a system for distributing file data that uses market forces to create competition between, and evolution of, many small agents that perform the functions needed to get the data from where it is created to were it is needed. The first step is creation of a suitably modular, decentralized and distributed file system architecture, with components which are sufficiently autonomous to allow their independent development, optimization and eventually evolution. Besides providing increased performance and efficiency such a system can greatly enhance the data security available to users.
Books:
On using market forces to optimize resource utilization. All three of the papers are absolutely fantastic. I give them my highest recommendation.
Brings together some very important ideas from a broad spectrum of disiplines. He reaches some interesting conclusions that will surprise almost everyone. It is one of the best books I've read in a long time.
This book is a bit different from the others in this category but still an important one. The title pretty much says it all.
I have been following ALife research for quite a while now, and this book manages to summarize some of the most important results of that research. In addition, Kauffman applies these ideas in interesting ways in several diverse realms. In evolutionary theory, he argues that natural selection doesn't work alone but depends crucially on natural self-organization. In economics, he uses his ideas to provide some deep understanding about why, and when, free markets work better than centrally planned ones. He claims that the same arguments suggest that Democracy, more than just a passing fad, is an application of the same principles to Politics. Read this book.
I have finally written a real review.
Kauffman has extended the ideas in this, and his other books, to suggest a possible fourth law of thermodynamics that applies to thermodynamically open systems.
Smolin argues that the Universe is intrisically hospitable to complex systems. Very thought provoking.
Other References:
Mailing list:
An early proposal.
I've updated these ideas further in a mail message titled "The Information Silk Road".
I wrote a sort of taxonomy of distributed file system functions which may elucidate the path between here and there. In response to a message from Adam Back on the Eternity mailing list I replied with more on the benefits of dividing the file system into the namespace and a delivery system. Wei Dai responded with some thoughts on the namespace and location servers, and I followed up to a more comprehensive outline of how I see data location and namespace consistency working. Stimulated by the FreeNet ideas I wrote an updated description of my ideas for managing a distributed namespace. Expanding on the FreeNet adaptive caching network, here is a proposal for a really large scale data location and distribution system.
Please feel free to share your comments on the subject of distributed file systems as complex adaptive systems in this forum, provided courtesy of Quick Topic (formerly Take It Offline).
This is important because it is the third most important ingredient in the information society that is forming, after computers and networks. The consequences of this change in society is little realized by most people. However, Tim May has been instrumental in working out these consequences. His Cyphernomicon lays out many of his ideas in this area. Or see this gentler introduction to Crypto-Anarchy. These issues are discussed on the Cypherpunks mailing list. The consequences for ecomonics will arrive a bit faster than the political ones, and will probably be the fundamental driver of the change. Bob Hettinga is the tireless promoter the Geodesic Economy. Especially fun are his rants.
A big problem facing the crucial speedy deployment of crypto is government opposition.
Books:
While the changes wrought by the information revolution will seem earth shattering, hard on its heels will be the nanotechnology revolution, which truly will be earth shattering. Nanotechnology was started by Eric Drexler. The Foresight Institute tries to track and promote its development.
Books:
This is the book that started it all. A very readable layman's introduction. If you don't know much about nanotechnology you owe yourself this glimpse of the future.
The tour-de-force technical analysis of all aspects of bringing nanotechnology to fruition. I wasn't able to get through more than about the first 150 pages before burning out. Amazing.
Considering all the powerful technological forces at work in our lives, where are we ultimately headed? What does we even mean? What about the technological singularity? The Extropians and Transhumanists consider these issues.
Development of space and moving civilization off the planet, into the
solar system and then colonizing the rest of the galaxy is a necessity.
Marshall Savage describes one approach in The
Millennial Project: Colonizing the Galaxy in Eight Easy Steps
.
Given developments in computers, networks, and nanotechnology, the
optimal techniques for moving into space are uncertain. Clearly, the
technology used by NASA is, at best, a stopgap measure.
More recently, the success of SpaceX, a commercial company, delivering supplies to the International Space Station is an important milestone.
Some related links:
Very miscellaneous links of interest.
Pitfalls of Object-Oriented Development.
Grid Namespace for Files.
I do believe that where there is a choice only between cowardice and violence, I would advise violence.
-- Mahatma Gandhi, Source: Mohandas K. Gandhi, Young India, August 11, 1920, from: Fischer, Louis ed., The Essential Gandhi, 1962, pp. 156-57. Via: Lucky Green's <shamrock at netcom dot com> .sig of 31-Jan-97
What we seek is not the overthrow of the government but a situation in which it gets lost in the shuffle.
-- Duncan Frissell <frissell at panix dot com>, Cypherpunks 29-Aug-96
Information is the oxygen of the modern age. It wafts across the electrified borders. Breezes of electronic beams blow through the Iron Curtain as if it were lace.
-- Ronald Reagan (speaking before the Institut de France on June 15, 1989)
This is my PGP public key. Here are the particulars:
KeyID: 1F4B4907, UserID: Ted Anderson <ota@polyonymo.us> Fingerprint: D347 B1D9 1C71 A209 39A5 0625 8CAC 50BC 1F4B 4907