Slashdot Mirror

if you look at the actual research page you'll get much more in-depth information about this, far more than the article.

The researchers say that all of the following are possible using off-the-shelf hardware:

• Passive attacks to decrypt traffic based on statistical
  analysis.
• Active attack to inject new traffic from unauthorized mobile
  stations, based on known plaintext.
• Active attacks to decrypt traffic, based on tricking the access point.
• Dictionary-building attack that, after analysis of about a day's
  worth of traffic, allows real-time automated decryption of all traffic.

It only takes 5 hours to collect enough information to mount a statistical attack! They also describe both passive and active attacks that are possible in some detail. This isnt something to shrug off - even a passive attack is potentially very damaging. And it's not exotic hardware - you can get a lot of mileage just out of your consumer hardware.

There's also a draft of the paper available from the research group.

Some information about their analysis is available.

Personally, I wasn't counting on WEP anyway, which is why I didn't bother buying the Lucent Gold cards. I just wish IPsec were more common, so that I wouldn't have to tunnel quite so much through ssh.

Of course, then there are unencrypted wireless networks like the ones at USENIX. Dug Song's presentation on dsniff was a big hit; look for the "Passwords Found on a Wireless Network" paper. (PostScript only, sorry.)

here.

PocketPC version here.

I can just imagine someone who didn't bother reading the Policy trying to change their screensaver to something other than ads (read: seti@home) and just finding it change back to the ads (if Juno bothers trying, that is).

Turn your PC off if you want to.

Of course, if you don't turn your PC off like they order you are 100% responsible for any money that has to be spent fixing problems that arise from the computer constantly running.

Would a Beowulf cluster speed up Seti@home? What would be the best implementation?

If there's a reason to build a small cluster, I'd say it'd be seti.

____________________

I'm actually part Cliff Stoll disciple (well, I agree with him on many points), so I'm definitely not of the mind that the presence of computers can bolster most areas of education. However, there's a few points I'd like to bring up:

1) I'm convinced that learning to program boosted my understanding of mathematics and problem solving in general. From programming, I learned what variables and formulas and functions were. I learned to break down problems into smaller problems.
I heard someone say "Honestly, I don't know how people who can't code get by" a couple of years ago. When I taught High School Algebra a year ago, I saw that some people could code mathematically and some people couldn't, and wondered if a programming course or two would have helped (there was, though, a Linux/Perl hacker who couldn't factor polynomials to save his life).

That said, most "computers in education" initiatives probably don't really include the idea of teaching programming to every student.

2) Having computers in schools probably
does help with... learning about computers. It would provide kids who don't have computers at home with a chance to become literate with them. Or more.

Case in point: me & some of my classmates. My family had (up until I was a senior in high school) a TI 99/4A. A cool machine, and I love Parsec, but not a paragon of modern computing, even in the late 80s. At school, we were fortunate enough to have (donated by Word Perfect, I beleive) an ICON m68020 based unix workstation. So at age 16 I was lucky enough to have access to something UNIXish and learn all sorts of good things. Another guy in my class, Matt, completely absorbed the OS and I believe was making $50,000 a year in the early 90s working as a sysadmin/consultant. Not bad. Several others in that class have gone onto cool things.

3) I worked for a company called the Waterford Institute for a bit. They create software to teach young kids fundamentals of reading and math. Coincidentally, my Mom works as a remedial reading teacher, and the school where she was working used their software. It was her feeling that the software actually really helped many of the kids. Each exercise was very carefully designed by educators/artists/techwriters/programmers together (I've never seen software so well speced and designed) and they usually quite entertaining.

--

Yes, but convergence also gave us a touch screen on our fridge with the intention of having us surf the net in our kitchen while we cook.

Do you have any idea how much I'd love that?

So I'm lazy, but printing those things out just annoys me...
--

I don't know where you think you learned this, but it's wrong. The whole point of business reply mail is that you only pay when it gets USED. Otherwise they'd just buy envelopes with pre-printed stamps.

The USPS' site is useless, but here's one site which explains.

The capture & compression can be done in real time, my system is an AMD K7 Duron 650Mhz running on a MSI motherboard with 64Megs of ram. I usually leave the audio uncompressed, at full PCM 44.1khz stereo. I also set DivX ;-) low motion codec at 1 sec keyframes, and 6000 (max) kbps. Average compression is 26:1, somewhere near 200 KB/sec. WIth 10Gigs free, it has more than 20 hours left for recording :) I think you could get even more if you also compress the audio, in .wma at 64kbps, but a little bit faster procesor could be needed. Ah yes, the harddisk is just a Maxtor 30G IDE drive, with UDMA enabled.

I also use a little free scheduler called "Windows Scheduler" to do the automated capture (it saves keystrokes), and virtualdub itself can stop the record after certain conditions are met (like, n minutes passed, or only n megs free on disk).

So yes, your VCR is obsolete already, get a decent CPU and TV Tuner, and have a lot of fun.

Oh, and hear this tip: do the capture at YUY2 (raw) so you can enable the "noise filter", anything from the default (17?) to below (left) should be okay. You will be amazed of the magic this does with old tapes or not good enough tv signal, then choose the compression at the "compression" menu option, so to be done in real time after the raw capture and filtering.

Of course install the DivX ;-) lossy codec, and the very useful free opensource huffyuv lossless codec, use the lossy one when you need a long recording time, and the lossless when you need quality above anything else. Same with PCM (raw) audio vs mp3/wma (lossy).

BTW: Could somebody with the knowledge please take a look at VirtualDub's and huffyuv's source code? Maybe it could be ported to BeOS and Linux, now that we have the DivX ;-) Deux source at hand it could be useful. I hope video4linux 2 is ready :)

--

This project should do rough curve fitting first (ie do quick calculations on client computers, to find a general idea of the curve) and then select areas of interest to farm out more signifigant data to client computers.

I think you miss the point of the clients in SETI@Home. The purpose of the clients is to do the pre-filtering, so that Berkeley has a manageable number of candidates for subsequent processing.

The only real curves they have to work with are the curves of event rate against detection threshold, and they had good ideas of those already; these are used to set the reporting thresholds for the clients. They will consider themselves very lucky if they find even one confirmed ET, so they are not trying to construct maps of ET density.

The original concept did assume restricting the work units to ones from the plane of our galaxy, as those would give the highest proportion of Sun-like stars, but that was when they didn't think they would have enough clients to process all the data.

Unless and until we have detected several ETs (who may then tell us where to look for the others!) we have no means of calibrating any rules we use to select good candidates.

"The Internet was supposed to be all about freedom."

NO! The Internet was all about connecting different Department of Defense computers together so that resources could be shared! That's it! End of story! All of this nonsense about the Internet being some kind of utopian Libertarian commune is Leftist mind-rot! "The Economist" tends to lean that way. Slashdot should have linked to an article more like http://socrates.berkeley.edu/~newman/chap2.html

You can listen to examples of vocoder audio on this web page.

*cough*

Ya buy a bunch of these things and reprogram them to say and do "interesting" things. Then you carefully repackage them, sneak them back into the store, and put them on the shelf. Imagine the fun with Grandma gives one to uncle Fred!!!

The Barbie Liberation Organization did this a while back by switching the voiceboxes of talking Barbie dolls with those of talking G.I. Joes and then putting them back into toy stores. Lot's of little boys and girls got more than they expected from Santa that year!

You pay for the storage you use in OceanStore. Read the paper, especially the part about Responsible Parties and the utility model. If you keep uploading random data into the system, you will end up with a large OceanStore bill at the end of the month.

Also, someone mentioned banding together with a bunch of friends and creating a little private OceanStore of their own. This is a great idea, and one that I (personally) am very fond of. Each of you give up two-thirds of your 200 GB disk (they'll be here in no time), and you get reliable, fault-tolerant, highly-available storage in return. In this case, if someone fills up the shared space, you and your other friends kick him out of the group.

Finally, you could take a MojoNation-type approach an introduce an arbitrary currency to pay each other for storage.

Sean

Our recently published paper, OceanStore: An Architecture for Global-Scale Persistent Storage describes the system in more detail and can be found on our publications page.

Sean

As one of the graduate students working on OceanStore, I should add a little to this discussion.

Your point about data management being more expensive than the storage itself is absolutely correct. OceanStore addresses this issue in several ways:

First, we use replication and coding algorithms to ensure the integrity and durability of data. Documents that are actively being written to are managed by a group of servers participating in a Byzantine fault-tolerant algorithm. This ensures that despite machine failure or compromise (of up to approximately a third of the machines), your data is safe from loss and corruption. It also provides availability, since from the algorithm's point of view a failing server and an unavailable server are the same.

Data that is not actively being written is stored in Erasure-coded form and spread across the system. A rate N Erasure code breaks an object into Nb pieces, where b is the number of blocks in the object. If any arbitrary b of these pieces can later be recovered, the entire document can be reproduced. For example, with a rate 2 Erasure code, a 1 MB document will be broken into a number of blocks totaling 2 MB in size, such that if any 1 MB of them can be recovered, the whole document can be reproduced. Since each block can be stored on a different server, this gives tremendous durability to data. It also takes nice advantage of the fact that storage is cheaper than the management of storage. I should also mention that we include algorithms which verify the integrity of the reconstructed data.

Second, OceanStore has an introspection system which manages the placement of data throughout the system. While replication and coding keep data safe, introspection moves data around for optimal locality. If your data is across the world from you, you may not care that it is correct or durable, since it takes so long to get at it. Introspection uses pattern recognition techniques to discover what data is important to you and move it or cache it near your current location. This removes the necessity of paying administrators to discover this information and move the data manually in order to improve the performance of the system.

Finally, in order to locate all of this constantly moving information, OceanStore employs a two-tier location system which provides fast access to nearby data and availability to far-away data.

Our recently published paper, OceanStore: An Architecture for Global-Scale Persistent Storage describes these issues in more detail and can be found on our publications page.

Sean

More information can be found here: http://oceanstore.cs.berkeley.edu.

just because it's faster than light, doesn't mean it's instantaneous (which is what "infinitely fast' means). if you were referring to this article, this is waves travelling ftl, in a specially prepared cesium gas. i'd wait until they can do this with physical stuff (ie. particles, not waves) before placing orders for hyperdrives.

in 1945 the internal combustion engine and supporting bits occupied about 1 cubic metre. 55 years later it still does - and i wouldn't say it's any cheaper, either. since at this stage, we have no idea just what a hyperdrive is, saying they're going to be cheap and plentiful currently falls into the realm of wishful thinking. again, i'm not placing any orders until i know i can afford one.

"the thing about aliens is, they're alien." (argh! i can't remember who i got this from...)
why would they have to be peaceful and secretive? just because we haven't seen/heard them doesn't mean they're not out there, pillaging everything smaller than, say, jupiter. if you're into such thing things, try Stephen Baxter's Space for an interesting novel that suggests why we're not knee-deep in aliens. (sorry, no link. but it's a recent release, so it shouldn't be too hard to find) (oh, and for what it's worth, i am looking...)

ok, maybe i shouldn't have mentioned air and water, but i think the dirt will be difficult.

on the whole, possible, but it requires too many things to turn out perfectly my liking...

--

just because it's faster than light, doesn't mean it's instantaneous (which is what "infinitely fast' means). if you were referring to this article, this is waves travelling ftl, in a specially prepared cesium gas. i'd wait until they can do this with physical stuff (ie. particles, not waves) before placing orders for hyperdrives.

in 1945 the internal combustion engine and supporting bits occupied about 1 cubic metre. 55 years later it still does - and i wouldn't say it's any cheaper, either. since at this stage, we have no idea just what a hyperdrive is, saying they're going to be cheap and plentiful currently falls into the realm of wishful thinking. again, i'm not placing any orders until i know i can afford one.

"the thing about aliens is, they're alien." (argh! i can't remember who i got this from...)
why would they have to be peaceful and secretive? just because we haven't seen/heard them doesn't mean they're not out there, pillaging everything smaller than, say, jupiter. if you're into such thing things, try Stephen Baxter's Space for an interesting novel that suggests why we're not knee-deep in aliens. (sorry, no link. but it's a recent release, so it shouldn't be too hard to find) (oh, and for what it's worth, i am looking...)

ok, maybe i shouldn't have mentioned air and water, but i think the dirt will be difficult.

on the whole, possible, but it requires too many things to turn out perfectly my liking...

--

and SETI is an incredibly worthless disinformation campaign run to keep Earth in the dark about the presence of intelligent life outside of our solar system

And distributed.net is incredibly worthless disinformation campaign run to keep Earth in the dark about the presence of unbreakable strong encryption, correct?

When you buy a PC today, the specs include memory and CPU and not much else. We're pretty close to the day where CPU won't matter for personal workstations [...] Memory is similarly stagnating; unless you work with Photoshop or its ilk, chances are you have no use for >256MB.

I don't know if I buy this. People are always going to find good ways to waste memory and CPU. Games, for example, are among the most hardware-intensive applications out there. And that's just consumers. On the pro side, you have to do compiling, 3D rendering, real-time video and audio manipulation, encryption/decryption and scientific analysis. Plus, web sites will continue to get more and more complicated as bandwidth and other various technologies improve.

Then there's the Search for ExtraTerrestrial Intelligence...

- Scott


------
Scott Stevenson

If you did your effects in Mpeg 4 structured audio instead of MIDI, you might get considerably more performance.

Why? Because there is considerable research in compiling MP4-SA to C and then running the native code, to get greater performance out of arbitrary effects, filters, etc.

More info is available here


Nicholas C Weaver
nweaver@cs.berkeley.edu

Huh? It would appear that linux-gnu systems have contributed an infinite ammount of computation to the project! 6) linux-gnu 12672647 NaN years 2147483647 hr 2147483647 min 0NaN sec Weird! heh. here

Team Slashdot on Seti@Home

I hit the SETI@Home page tonight and discovered that they've upgraded the client _again_.

I was reading the note about their new client, which basically says that they've made the client slower.

Seems like they're running out of bandwidth on their server, so they've made the client do more research per packet in order to increase turnaround time, hence decreasing the rate that results are turned in.

It just seemed ironic in view of the 500,000 year announcement; SETI@Home has gotten too big for Berkeley's network.

Steve Magruder

• Asteroid impact... this has been a favorite since Alvarez found the Iridium layer. The theory goes that ~65 million years ago, a massive asteroid impacted, somewhere around the Carribean. From there, the theory fragments; some go for nuclear winter, others go for acid rain and worldwide forest-fires. There's lots of evidence that there was in fact an asteroid impact, but there's also a lot of evidence that the results were nowhere near as bad as everyone claims. Soil samples don't suggest acid rain, nor ice age. Most importantly, the fossil record doesn't support this. What everyone forgets is that the dinosaurs didn't die in a week. Think about it, if an asteroid did all the work, you'd expect there to be a colossal slaughter, but we don't see that in the record...

• Plants became poisonous and killed off dinosaurs... too many issues with this theory.

• Mass migration moved diseases around which killed the dinos. This wouldn't be enough to remove them all, but it's got some good points: it would target large mobile critters more, and a disease would be more likely to affect similar animals (like dinos) but exclude others. Still, there are issues here...

• Deccan Traps. When India slammed into the Asian continent, it pushed up the Himalayas, and created lava flows that covered millions of square acres. The amount of ejecta from the eruptions could have created a nuclear winter on its own (think St. Helens but millions of times worse).

Most likely it was a combination of many of these things... Diseases start to ravage certain lineages, others can't compete with new types of plants, and slowly die off. The climate changes due to massive volcanic eruptions, and this adds more challenges to various species. After a few million years of this, dinosaurs as a class are hurting: they have little species diversity left, and are unlikely to recover. An asteroid impact puts the final touch on it...

For more reading check the following links:

A Cowen Essay posted on UCMP
Summaries from UCMP
Also, see The Dinosaur Heresies by Robert Bakker... His ideas are interesting and opposite of much of UCMP's thoughts. But he's definitely worth reading; I recommend this to anyone interested in dinos...

• Asteroid impact... this has been a favorite since Alvarez found the Iridium layer. The theory goes that ~65 million years ago, a massive asteroid impacted, somewhere around the Carribean. From there, the theory fragments; some go for nuclear winter, others go for acid rain and worldwide forest-fires. There's lots of evidence that there was in fact an asteroid impact, but there's also a lot of evidence that the results were nowhere near as bad as everyone claims. Soil samples don't suggest acid rain, nor ice age. Most importantly, the fossil record doesn't support this. What everyone forgets is that the dinosaurs didn't die in a week. Think about it, if an asteroid did all the work, you'd expect there to be a colossal slaughter, but we don't see that in the record...

• Plants became poisonous and killed off dinosaurs... too many issues with this theory.

• Mass migration moved diseases around which killed the dinos. This wouldn't be enough to remove them all, but it's got some good points: it would target large mobile critters more, and a disease would be more likely to affect similar animals (like dinos) but exclude others. Still, there are issues here...

• Deccan Traps. When India slammed into the Asian continent, it pushed up the Himalayas, and created lava flows that covered millions of square acres. The amount of ejecta from the eruptions could have created a nuclear winter on its own (think St. Helens but millions of times worse).

Most likely it was a combination of many of these things... Diseases start to ravage certain lineages, others can't compete with new types of plants, and slowly die off. The climate changes due to massive volcanic eruptions, and this adds more challenges to various species. After a few million years of this, dinosaurs as a class are hurting: they have little species diversity left, and are unlikely to recover. An asteroid impact puts the final touch on it...

For more reading check the following links:

A Cowen Essay posted on UCMP
Summaries from UCMP
Also, see The Dinosaur Heresies by Robert Bakker... His ideas are interesting and opposite of much of UCMP's thoughts. But he's definitely worth reading; I recommend this to anyone interested in dinos...

Checking out Seti's results by OS shows some interesting info. In the 90 OSes listed, MacOS is #3 in results and OS/2 is #20. Linux is #6. Based on SETI's results, F@H should have done the Mac client BEFORE the linux client.

Wow.

-S

Also fun is the Countries "Sorted by Most CPU Time per capita":
Link
USA at number 12 behinda Antarctica, Pitcairn Islands, Tokelau, Niue (wtf?) and other assorted specs of fly-shit on the world map

Look at the results per CPU/OS...
5th entry ;-)
--

I have this great idea for a honeypot, although it might seem a little futuristic.

Picture this: we create a series of directories that contain apparently classified military information. We'll call it something obscure, some sort of acronymn, like SDINet, for example . . . I bet that would keep a dedicated hacker occupied for hours, especially if you mixed in some binary files so they had to check each one before trying to view it on the server.

I know it seems bizzare, but I think it actually might work! And the best part is I don't think anyone has ever come up with anything like this before!

Let me know if you think it would work?

--

The article on the Intel transistor seems to refer to a 30nm thickness (I assume they mean the gate dielectric). But when most people talk about the "size" of a transistor, they refer to the gate length. As far as I know, the 18nm gate length transistors developed at UC Berkeley (and discussed in this Slashdot article) are the "smallest."

-Jason

The article on the Intel transistor seems to refer to a 30nm thickness (I assume they mean the gate dielectric). But when most people talk about the "size" of a transistor, they refer to the gate length. As far as I know, the 18nm gate length transistors developed at UC Berkeley (and discussed in this Slashdot article) are the "smallest."

-Jason

From the FAQ

Sather is an object oriented language designed to be simple, efficient, safe, flexible and non-proprietary. One way of placing it in the "space of languages" is to say that it aims to be:

• As efficient as C, C++, or Fortran
• As elegant as and safer than Eiffel
• Support higher-order functions and iteration abstraction similar to Common Lisp, Scheme, CLU, or Smalltalk.

Sather has garbage collection, statically-checked strong typing, multiple inheritance, separate implementation and type inheritance, parameterized classes, dynamic dispatch, iteration abstraction, higher-order routines and iters, exception handling, assertions, preconditions, postconditions, and class invariants. Sather code can be compiled into C code and can efficiently link with C object files. The upcoming version of Sather will have full internationalization support.

class MAIN is
main is
#OUT + "Hello World!\n";
end; -- method main
end; -- class MAIN

• GNU Sather
• Sather Home Page
• FAQ

No, Jon, that was not a study of Sims Users, but a study of the amount of information the Internet carries by the School of Information Management and Systems at Berkeley. Geez, John, you could try actually following the link and backtracking to find the actual source of something....

No, Jon, that was not a study of Sims Users, but a study of the amount of information the Internet carries by the School of Information Management and Systems at Berkeley. Geez, John, you could try actually following the link and backtracking to find the actual source of something....

Unfortunately, I can't find a link for Dr. Hankopf's paper, but it primarily focused on the necessity of some centralized control of mains electrical power, if a technological society is to flourish. Hankopf claims that free power is merely

'mythopoetical libertarianism,' which imposes upon a newly-classless society the semiotics of a sadly posttextual paradigm. But an abundance of narratives concerning the economy, and subsequent stasis, of cultural society may be found. A Marxist approach will destroy the semiotics of a technology culture and, ultimately, cause its collapse.

There is no need to dissect Hankopf's bias -- as a card-carrying Objectivist, she clearly feels that corporations with a profit motive will provide the best foundation for a modern society. However, she raises some interesting points, especially in light of the Oracle debacle.

Most of the top rated systems throughout the world, sending packets for SETI@Home, are Compaq servers running Tru64 Unix. Most of this is due to the scientific data using 64bit accuracy, for which the "contemporary" systems of 32 bits just aren't adequate.

Other applications that crush with 64 bits include high-quality graphic rendering, vast database addressing, and, oh yeah, NETSCAPE 6! ;-)

It is not a reflection on the XCF's current membership, but it is important to understand that the XCF was created by destroying a public (to the student body at least) free student-run computing facility, the Undergraduate Computing Facility. The equipment and space was essentially given to XCF in the belief that they would continue the UCF, but instead the founders created a private facility for themselves.

As a result, for more than a year, there was no public access computing for undergraduates, and it took a long time and a lot of work to rebuild what XCF co-opted, eventually in the form of UC Berkeley's Open Computing Facility.

Over the past fifteen years, some significant software was written at XCF, especially after they started releasing source code. And, certainly, eliminting the "Exclusive Computing Facility", as we called it, rather than just making it more open, would not have helped anyone. However, there is no doubt in my mind that for a school as big as Berkeley with such a long history of system software development, that more and better development would have been done had XCF been a more open facility from the outset, and it would have defined the campus programming community in a much more postive way.

By now, surely, none of the current XCF members had anything to do with its bloody founding, fifteen years ago. There has been public computing at Cal for a long time (at least as of the last time I checked), and multitasking computers are ubiquitous. So, the significance of the XCF to me now is only symbolic. However, as an alumnus, I would somehow feel better about about Cal to see the experimental computing club reconstituted into something that does not claim to be the legacy of the founding of the XCF. The current members do not deserve the dishonor of that association.

It is expected to translate 80x86 instructions into VLIW instructions (or directly to decoded instructions) the same way that Intel P6 and AMD K5/K6/K7 CPUs do, but with a larger number of instructions issued using the VLIW design, it should be faster. However, if native IA-64 code is even faster, this may finally produce the incentive to let the 80x86 architecture finally fade away.

scary

better have that 4000 way Itty workstation to run my 16 bit apps

Burris

Does anyone know if there is an actual production Virtual Keyboard being made? I did a quick google search and found these guys at Berkeley, but this looks like a design project. I'm getting worried about Carpal tunnel and arthritis and would like to do something other than bang on a keyboard all day. A keyboard glove would seem to be a solution since you are not making contact with keys, but just typing in the air.

first they take all my tax dollars and give them to some longhair running a screen saver so that they can scan for alien signals
I think SETI@Home is a privately funded group. Now your government might be pumping money into one or more of the corporate sponsors found on the main page, but that is completely up to the corporates.
---

The Crusoe's not really aimed at the PDA market. It's way too overpowered and as energy stingy as it is, it doesn't come anywhere near todays PDA processors.

The Crusoe TM 3200 - 333mhz (Transmeta's most "mobile" processor) consumes ~15mW of power in it's most power conservative state (deep sleep idle) and averages 1.4W (mp3 playback) (both of these figures include the Northbridge chipset power consumption as well).
By comparison, an entire Palm Pilot Pro (using a 16mhz Dragonball eats ~26mW in it's idle state and consumes ~160mW when running CPU intensive applications.

While you could leave your Crusoe on the shelf and outlast a Palm, It wouldnt take very long for 1.4W to drain a pair of AAA batteries once you start doing something with the unit.

The Crusoe's not really aimed at the PDA market. It's way too overpowered and as energy stingy as it is, it doesn't come anywhere near todays PDA processors.

The Crusoe TM 3200 - 333mhz (Transmeta's most "mobile" processor) consumes ~15mW of power in it's most power conservative state (deep sleep idle) and averages 1.4W (mp3 playback) (both of these figures include the Northbridge chipset power consumption as well).
By comparison, an entire Palm Pilot Pro (using a 16mhz Dragonball eats ~26mW in it's idle state and consumes ~160mW when running CPU intensive applications.

While you could leave your Crusoe on the shelf and outlast a Palm, It wouldnt take very long for 1.4W to drain a pair of AAA batteries once you start doing something with the unit.

....go to the IMAGE (Imager for Magnetopause-to-Aurora Global Exploration) "WIC image" site at: http://sprg.ssl.berkeley.edu/im age /latest_wic.html. The pictures of the nothern hemisphere aurora are updated every few minutes directly from the Far Ultra Violet (FUV) Instrument aboard the sattelite.

Basically, there are two ways to preform replication: "lazy replication" and "eager replication". "Eager replication" means that all updates are atomic across all nodes and that transactions are serializable. However, the problem with "eager replication" is that as you increase the number of nodes n, the probability of deadlock increases on the order of n^5. The "solution", such as it is, is to remove the expectation of serializabilty, using timestamps for concurrency control, only allow commutative transactions on your data, and use two-tier replication. This works for banks and others whose database applications consist mainly of commutative transactions, but won't for many others: YMMV. (Gray's paper also details the differences between having a single "master" node that "owns" all db objects and having each node own several objects.)

IIRC, the way Notes does it is by queuing updates at the local node and using an optimistic concurrency control mechanism when the local node connects to replicate. This is great for the application domain that Notes caters to: I "own" my own calendar, and if I'm out of the office (and have my node -- notebook -- with me), you can't schedule me for an appointment until I come back. However, for many application domains, this won't work.

In any case, that's why Notes does it -- because it can, thanks to the nature of its data domain -- and why most people don't -- because it's hard/impossible for the general case.


~wog
PS -- If you can't get into the ACM Digital Library, check out these lecture notes from Stonebraker's anthology at Berzerkeley.