Both researchers and operators of production systems are frequently faced with the need to manipulate entire disk images. Convenient and fast tools for saving, transferring, and installing entire disk images make disaster recovery, operating system installation, and many other tasks significantly easier. In a research environment, making such tools available to users greatly encourages experimentation.
We present Frisbee, a system for saving, transferring, and installing entire disk images, whose goals are speed and scalability in a LAN environment. Among the techniques Frisbee uses are an appropriately-adapted method of filesystem-aware compression, a custom applicationlevel reliable multicast protocol, and flexible applicationlevel framing. This design results in a system which can rapidly and reliably distribute a disk image to many clients simultaneously. For example, Frisbee can write a total of 50 gigabytes of data to 80 disks in 34 seconds on commodity PC hardware. We describe Frisbees design and implementation, review important design decisions, and evaluate its performance.
I think the usual practice in america is more like a couple thousand dollars and a shiny brass plaque.
I wouldn't lose sleep over the bonus. Instead, remember to mention your patent at your next performance review. Even if you don't get a bonus from it directly it may be a useful bargaining chip for future compensation.
Re:at the limit it actually would be a good thing.
on
Building Better Spam
·
· Score: 1
Not so fast. It is my understanding that telemarketers and spammers often exploit those are not making rational decisions for themselves and are unable to say no. This is the legal version of taking candy from a baby.
Just because it is happening to someone else's addled grandmother or retarded cousin, shouldn't absolve any of us from our responsibility to protect the weak (and kill spammers dead).
There are already fair trade practices laws that cover these cases and these people still get spam. I was mostly trying to make the point that sending an unsolicited commercial email is not in itself an evil act. If I'm trying to sell my old bike and I hear through a mutual friend that you're looking for a cheap bike then my unsolicited commercial email to you might be quite welcome. If that was the only kind of spam you ever recieved it wouldn't be an issue.
If you recieved say 2-3 spams a day for products you might concievably want then the spam crisis such as it is would be over. Sure there would still be some people trying to kill spam dead but just reducing it to a manageable amount would be good enough for 99% of people. The weekly supermarket flyer I get in the snail mail usually goes right in the bin but I'm not too mad about it because I don't get buried in them and I might at least potentially be interested in this weeks sales.
Unfortunately this kind of targetted spam won't happen until we pass a law requiring users to register their penis size with their ISP.
at the limit it actually would be a good thing...
on
Building Better Spam
·
· Score: 3, Insightful
If everyone was committed to hitting a high enough rate of return with spam that could be great with users. Imagine spam with a 100% success rate. That would mean it was only mailed to people who actually wanted to buy the product or service. I'd say that would be a win for everyone.
Remember how back in the early days of internet advertising the starry eyed utopians talked about how you could use advanced techniques to send email advertisements only to those who were probably interested? Of course these were utopians we're talking about so they didn't bother doing even a back of the envelope calculation of the cost of finding the right 300 people to send your ad to versus just sending it to 10 million.
Unfortunately my understanding is the software referenced in cringely's article doesn't find the "right" people to spam, it just helps you punch up your ad copy. Which might double a spammers response rate from.0002% to.0004%. Still not enough to cut down on the amount of spam we recieve.
The poster also slightly misrepresents cringely's article since cringely's not advocating the use of the software for spam but rather for auction listings.
I'm pretty sure "that Doug guy" worked for SRI, the Stanford Research Institute, in Menlo Park (SRI not being affiliated with Stanford; Menlo Park, CA not being affiliated with Menlo Park, NJ (home of Edison)). I understand that back in the day there was a lot of cross-pollination between PARC, SRI, and Stanford though.
The problem with that is it denies the possibility of true international law.
And how is that a problem? Right now supreme power rests with the nation-state. That is the order of things. Perhaps this is changing, but not to the extent that nations allow organizations like the UN to push them around. Iraq and Israel would be two example of this.
It's a problem because diplomacy tends to favor the strong. When two nations sit down at a bargaining table they don't sit down as equals. When two nations argue in court before a judge there is at least a pretension of equality, based in judicial traditions.
The ABM Treaty alone is a crucial factor in national security; letting Bush get away with facilitating its demise will destroy the balance of powers carefully crafted in our Constitution. - is it really their position that a treaty supercedes US law, and an act of Congress is needed to repudiate it? Has any previous government in US history been challenged in a US court for breaking a treaty? That comes straight out of the old right-wing paranoids' one-world-government book.
Since you asked, from the United States Constitution:
Article II, Section 2, Clause 2: [The President] shall have Power, by and with the Advice and Consent of the Senate, to make Treaties, provided two thirds of the Senators present concur
Article VI, Clause 2: This Constitution, and the Laws of the United States which shall be made in Pursuance thereof; and all Treaties made, or which shall be made, under the Authority of the United States, shall be the supreme Law of the Land; and the Judges in every State shall be bound thereby, any Thing in the Constitution or Laws of any State to the Contrary notwithstanding.
So it looks like treaties have the same standing as US laws. In this particular case I think the ABM treaty has a withdrawal clause, so maybe it doesn't require an act of congress to repudiate it, but in general I think it would take an act of congress to repudiate a treaty based on the clauses I quoted above.
I don't know about the court records but I think if the president broke a treaty I think he would end up in court. The constitution recognizes the US Supreme Court as the only court competent to hear such a case, which is probably why so many Americans (conservatives especially I guess) don't like the World Court in the Hague. I almost see their point, but it seems a little self-centered. What happens when two nations have a disagreement where the recognize their own supreme court as the only competent jurisdiction? I think the practical answer is that these issues have always been worked out diplomatically in the past and could continue to be worked out so. The problem with that is it denies the possibility of true international law. Which is fine if you can always negotiate from a position of strength (as the US can now) but leaves the little guy in the lurch. Which is one of the main reasons for laws in the first place, to protect the weak.
My understanding is that registration isn't required in order for your work to be copyrighted, and hasn't been required since at least 1976. Everything I read on this give some line about how registering a copyright makes your court case easier if you have so sue someone over infringement, but I wonder how many published works are registered.
I would venture to guess that most mainstream works are.
It's worse than that. Even if none of the people receiving spam are suckers enough to buy the products advertised, there are still plenty of suckers on the sending end, believing that spamming millions of people is going to make them rich, or at least recoup their investment.
I don't think SMTP is really the problem. What feature of a new protocol would prevent spam while still allowing all legitimate mail through. Why does that feature require a whole new protocol?
Pull the hard drive, then ship it UPS (or Fedex, or UPS) insured. About 50% of the time they'll trash your computer but they'll give you cash equal to what *you paid* for your computer, not what it's worth today. Think of it as regular free upgrades.
As long as you don't trust them with your data this is a much easier policy than trying to figure out a safe way to ship it.
As far as packaging, I've had good luck double boxing it (put computer inside tight fitting box packed with foam, put box inside larger looser box packed with styrofoam peanuts). The outside box ends up looking trashed but the inside box is fine. Again this is with pulling the harddrives and carrying those personally.
Oh, ATM's used quite a bit. It just sucks. Let's take DSL for example. DSL modems encapsulate ethernet frames in ATM cells. The ATM cell size is 48 data bytes (plus a header, 8 bytes IIRC).
Take the shortest common datagram, at TCP ACK. That's typically 20 bytes of IP header + 20 bytes of TCP header with no options or data bytes. Put that in an ethernet frame. It's padded to 64 data bytes (minimum frame length), plus 16 byte header plus 2 byte trailer (CRC). That's 82 bytes.
So for a packet with no data you have 2 ATM cells which are fixed at 48 bytes long. Plus cell headers that's 112 bytes for a 40 byte packet of no data.
ATM gear is quite expensive since it supports QoS and reliable delivery at the cell level. Ironically when running TCP/IP over ATM (with or without an ethernet encapsulation or emulation layer) this all goes wasted since TCP/IP will do that for you anyway. And for applications which don't need those guarantees it essentially means added latency and reduced bandwidth for no value.
Now the small uniform cell size does have some benefits, especially in a streaming mode like voice traffic.
Uhhh... You can run ethernet over fibre. It works quite well.
The problem with Fibre Channel is that the storage people are having to recreate all of networking, and make all of the early networking mistakes all over again. Painful. That and it's expensive.
I can't speak to SONET but if it's the disaster that ATM is/was then there's not a lot to learn there. Maybe some QoS stuff and a lot of bad examples.
For the record ATM and Fibre Channel both commonly run over copper as well as fibre. The physical medium for all of these things makes a lot less difference than you seem to think.
I haven't read the SciAm article so I'm not sure what spin they put on it, but it's actually a very reasonable idea.
The idea is two-fold:
(a) When trying to maximize reliability, it might actually be better in terms of total downtime to reduce recovery time rather than improve reliability. Take a system which crashes 5 times a year on average and takes an hour to go back online each time it crashes. Your total downtime is 5 hours/year. If you fix one place where the system crashes your total downtime will go down to 4 hours/year. But maybe for the same effort you can reduce the recovery time from 1 hour to 45 minutes. That's 3.75 hours/year of downtime. This is the kind of tradeoff that a lot of reliability engineering people don't think about, but should.
At the limit, if you had a file server that could recover within 5 seconds, who cares if it crashes twice a day? That's a short enough interval that the clients will automatically retry and succeed.
(b) You have to design the recovery path anyway, since you have to assume that sometimes your system crashes. You could also design a clean shutdown / startup path OR you could put all of that effort into making your recovery path that much faster and more effective.
Not having a "clean" shutdown path also has the benefit that every time you restart the system for any reason you are testing your recovery logic.
Actually, as noted in the documentation of early versions of LHarc, it's conceptually derived from LZari, another archiving program from Japan. There were lots of different archivers back in the day that largely shared a command set, including the early arc and zoo and the later arj. Some people may better remember lharc as `lha' as it was later renamed to match its file extension.
I don't know whether Yoshi's or Katz's program came first, but I find it unlikely that one was a "knockoff" of the other.
Geez. I haven't used LHarc in 10 years. I can't believe I still remember this junk.
Unless (or until) your business becomes a public nuisance your local government will probably be very supportive. Since they will be collecting tax revenue on your business (directly or indirectly) it's in their best interests for you to succeed.
Larger cities often have offices whose sole purpose is to help small businesses. They are probably your best first resource for understanding basic legal issues related to your business (i.e. what taxes you have to pay, what permits are required, etc.). The people in that office can be a great resource for lots of other questions as well since they see people like you every day.
Your local chamber of commerce may be a good resource too. When I was starting a small web development business a few years ago they seemed mostly interested in collecting membership dues though, so YMMV.
It's been a while since I've needed to know this for anything, but...
There are two page sizes on IA32, 4k and 4m. Actually I think there's a mode with 2m pages as well, but we can lump it in with the 4m case for our purposes.
Windows uses 4m pages for at least some always-resident kernel pages. Linux I believe always uses 4k pages but I'm not certain. There are issues with 4m pages, like they're annoying for copy-on-write or swapping out. Also less convenient for page-on-demand for executables. In fact, I would venture to say that they're only useful in rather specialized situations. It's not that often that you have 4m of pinned non-CoW memory (maybe video memory would be the only case in most systems).
As another poster rightly pointed out we should worry too much about a practical limit of 16 gigs, since there's not really a market demand for this in IA32. When it gets to be an issue somebody will find a way around it.
Of the people who might want >16 gigs of RAM right now I expect almost all of them are working with large databases of one form or another and could work around the hassles of 4m pages, thus scaling all the way to 64 gig.
While theoretically you can put up to 64 gigs into a recent IA32 machine, my understanding is that in Linux at least the practical limit is 16 gigabytes. After that the page tables won't fit in kernel space.
I'd expect that most other OSes have limits like that due to architectures designed when nobody was close to using a full 4 gigs.
Actually I don't know that the majority of Xfree86 users do not use remote X. I can't think of anyone I've ever known who uses X and doesn't use the remoting feature at least occasionally.
For my part, it's almost always the case that the only applications that are local are my web browser and my mail client.
If you are part of an organization that has a Unix infrastucture (such as a university or many computer industry businesses) and use Unix yourself, you are almost certainly a user of remote X.
I guess my point would be that if you have more than one Unix machine you probably use X remoting at least some of the time. Unless maybe you're from a Windows/Mac background and just aren't used to the convenience of X.
While we're at it I think Thompson owns RCA. RCA is these days just a brand of some European company in any case.
Re:Other ways the market should be working
on
LCD Price Fixing?
·
· Score: 1
My laptop has a single dead pixel, or more accurately 1/3 of a dead pixel. It will show green and blue but not red. It's slightly annoying but you get used to it.
If you really want to stay within the bounds of science, the principle of parsimony only applies when you actually have a theory. "I don't know how the universe was created" isn't a theory, so you can't eliminate "God created the universe" on the grounds that the first is simpler.
By way of analogy, suppose I don't know how electricity. My statement to that effect is more parsimonious than your explanation that the phenomenon we observe as electricity is the flow of electrons, but that doesn't matter because in this case you have a theory and I don't.
It's also important to understand that science does not purport to have THE answer. It purports to have AN answer, and in particular a useful answer. The principles of science prefer certain kinds of answers to others (falsifiable, parsimonious, etc.)
Slashdot Mirror
From the abstract:
http://www.cs.utah.edu/flux/papers/frisbee-usen
I think the usual practice in america is more like a couple thousand dollars and a shiny brass plaque.
I wouldn't lose sleep over the bonus. Instead, remember to mention your patent at your next performance review. Even if you don't get a bonus from it directly it may be a useful bargaining chip for future compensation.
Not so fast. It is my understanding that telemarketers and spammers often exploit those are not making rational decisions for themselves and are unable to say no. This is the legal version of taking candy from a baby.
Just because it is happening to someone else's addled grandmother or retarded cousin, shouldn't absolve any of us from our responsibility to protect the weak (and kill spammers dead).
There are already fair trade practices laws that cover these cases and these people still get spam. I was mostly trying to make the point that sending an unsolicited commercial email is not in itself an evil act. If I'm trying to sell my old bike and I hear through a mutual friend that you're looking for a cheap bike then my unsolicited commercial email to you might be quite welcome. If that was the only kind of spam you ever recieved it wouldn't be an issue.
If you recieved say 2-3 spams a day for products you might concievably want then the spam crisis such as it is would be over. Sure there would still be some people trying to kill spam dead but just reducing it to a manageable amount would be good enough for 99% of people. The weekly supermarket flyer I get in the snail mail usually goes right in the bin but I'm not too mad about it because I don't get buried in them and I might at least potentially be interested in this weeks sales.
Unfortunately this kind of targetted spam won't happen until we pass a law requiring users to register their penis size with their ISP.
If everyone was committed to hitting a high enough rate of return with spam that could be great with users. Imagine spam with a 100% success rate. That would mean it was only mailed to people who actually wanted to buy the product or service. I'd say that would be a win for everyone.
.0002% to .0004%. Still not enough to cut down on the amount of spam we recieve.
Remember how back in the early days of internet advertising the starry eyed utopians talked about how you could use advanced techniques to send email advertisements only to those who were probably interested? Of course these were utopians we're talking about so they didn't bother doing even a back of the envelope calculation of the cost of finding the right 300 people to send your ad to versus just sending it to 10 million.
Unfortunately my understanding is the software referenced in cringely's article doesn't find the "right" people to spam, it just helps you punch up your ad copy. Which might double a spammers response rate from
The poster also slightly misrepresents cringely's article since cringely's not advocating the use of the software for spam but rather for auction listings.
I'm pretty sure "that Doug guy" worked for SRI, the Stanford Research Institute, in Menlo Park (SRI not being affiliated with Stanford; Menlo Park, CA not being affiliated with Menlo Park, NJ (home of Edison)). I understand that back in the day there was a lot of cross-pollination between PARC, SRI, and Stanford though.
It's a problem because diplomacy tends to favor the strong. When two nations sit down at a bargaining table they don't sit down as equals. When two nations argue in court before a judge there is at least a pretension of equality, based in judicial traditions.
Since you asked, from the United States Constitution:
So it looks like treaties have the same standing as US laws. In this particular case I think the ABM treaty has a withdrawal clause, so maybe it doesn't require an act of congress to repudiate it, but in general I think it would take an act of congress to repudiate a treaty based on the clauses I quoted above.
I don't know about the court records but I think if the president broke a treaty I think he would end up in court. The constitution recognizes the US Supreme Court as the only court competent to hear such a case, which is probably why so many Americans (conservatives especially I guess) don't like the World Court in the Hague. I almost see their point, but it seems a little self-centered. What happens when two nations have a disagreement where the recognize their own supreme court as the only competent jurisdiction? I think the practical answer is that these issues have always been worked out diplomatically in the past and could continue to be worked out so. The problem with that is it denies the possibility of true international law. Which is fine if you can always negotiate from a position of strength (as the US can now) but leaves the little guy in the lurch. Which is one of the main reasons for laws in the first place, to protect the weak.
My understanding is that registration isn't required in order for your work to be copyrighted, and hasn't been required since at least 1976. Everything I read on this give some line about how registering a copyright makes your court case easier if you have so sue someone over infringement, but I wonder how many published works are registered.
I would venture to guess that most mainstream works are.
Biology for one. Biologists don't "do" computers, for the most part.
It's worse than that. Even if none of the people receiving spam are suckers enough to buy the products advertised, there are still plenty of suckers on the sending end, believing that spamming millions of people is going to make them rich, or at least recoup their investment.
I don't think SMTP is really the problem. What feature of a new protocol would prevent spam while still allowing all legitimate mail through. Why does that feature require a whole new protocol?
Pull the hard drive, then ship it UPS (or Fedex, or UPS) insured. About 50% of the time they'll trash your computer but they'll give you cash equal to what *you paid* for your computer, not what it's worth today. Think of it as regular free upgrades.
As long as you don't trust them with your data this is a much easier policy than trying to figure out a safe way to ship it.
As far as packaging, I've had good luck double boxing it (put computer inside tight fitting box packed with foam, put box inside larger looser box packed with styrofoam peanuts). The outside box ends up looking trashed but the inside box is fine. Again this is with pulling the harddrives and carrying those personally.
It's quite possible that Sun negotiated a separate license from Id, in which case the GPL wouldn't apply.
Oh, ATM's used quite a bit. It just sucks. Let's take DSL for example. DSL modems encapsulate ethernet frames in ATM cells. The ATM cell size is 48 data bytes (plus a header, 8 bytes IIRC).
Take the shortest common datagram, at TCP ACK. That's typically 20 bytes of IP header + 20 bytes of TCP header with no options or data bytes. Put that in an ethernet frame. It's padded to 64 data bytes (minimum frame length), plus 16 byte header plus 2 byte trailer (CRC). That's 82 bytes.
So for a packet with no data you have 2 ATM cells which are fixed at 48 bytes long. Plus cell headers that's 112 bytes for a 40 byte packet of no data.
ATM gear is quite expensive since it supports QoS and reliable delivery at the cell level. Ironically when running TCP/IP over ATM (with or without an ethernet encapsulation or emulation layer) this all goes wasted since TCP/IP will do that for you anyway. And for applications which don't need those guarantees it essentially means added latency and reduced bandwidth for no value.
Now the small uniform cell size does have some benefits, especially in a streaming mode like voice traffic.
Uhhh... You can run ethernet over fibre. It works quite well.
The problem with Fibre Channel is that the storage people are having to recreate all of networking, and make all of the early networking mistakes all over again. Painful. That and it's expensive.
I can't speak to SONET but if it's the disaster that ATM is/was then there's not a lot to learn there. Maybe some QoS stuff and a lot of bad examples.
For the record ATM and Fibre Channel both commonly run over copper as well as fibre. The physical medium for all of these things makes a lot less difference than you seem to think.
http://roc.cs.berkeley.edu/
I haven't read the SciAm article so I'm not sure what spin they put on it, but it's actually a very reasonable idea.
The idea is two-fold:
(a) When trying to maximize reliability, it might actually be better in terms of total downtime to reduce recovery time rather than improve reliability. Take a system which crashes 5 times a year on average and takes an hour to go back online each time it crashes. Your total downtime is 5 hours/year. If you fix one place where the system crashes your total downtime will go down to 4 hours/year. But maybe for the same effort you can reduce the recovery time from 1 hour to 45 minutes. That's 3.75 hours/year of downtime. This is the kind of tradeoff that a lot of reliability engineering people don't think about, but should.
At the limit, if you had a file server that could recover within 5 seconds, who cares if it crashes twice a day? That's a short enough interval that the clients will automatically retry and succeed.
(b) You have to design the recovery path anyway, since you have to assume that sometimes your system crashes. You could also design a clean shutdown / startup path OR you could put all of that effort into making your recovery path that much faster and more effective.
Not having a "clean" shutdown path also has the benefit that every time you restart the system for any reason you are testing your recovery logic.
Actually, as noted in the documentation of early versions of LHarc, it's conceptually derived from LZari, another archiving program from Japan. There were lots of different archivers back in the day that largely shared a command set, including the early arc and zoo and the later arj. Some people may better remember lharc as `lha' as it was later renamed to match its file extension.
I don't know whether Yoshi's or Katz's program came first, but I find it unlikely that one was a "knockoff" of the other.
Geez. I haven't used LHarc in 10 years. I can't believe I still remember this junk.
I think it's a typo. He actually meant to say:
Unless (or until) your business becomes a public nuisance your local government will probably be very supportive. Since they will be collecting tax revenue on your business (directly or indirectly) it's in their best interests for you to succeed.
Larger cities often have offices whose sole purpose is to help small businesses. They are probably your best first resource for understanding basic legal issues related to your business (i.e. what taxes you have to pay, what permits are required, etc.). The people in that office can be a great resource for lots of other questions as well since they see people like you every day.
Your local chamber of commerce may be a good resource too. When I was starting a small web development business a few years ago they seemed mostly interested in collecting membership dues though, so YMMV.
It's been a while since I've needed to know this for anything, but...
There are two page sizes on IA32, 4k and 4m. Actually I think there's a mode with 2m pages as well, but we can lump it in with the 4m case for our purposes.
Windows uses 4m pages for at least some always-resident kernel pages. Linux I believe always uses 4k pages but I'm not certain. There are issues with 4m pages, like they're annoying for copy-on-write or swapping out. Also less convenient for page-on-demand for executables. In fact, I would venture to say that they're only useful in rather specialized situations. It's not that often that you have 4m of pinned non-CoW memory (maybe video memory would be the only case in most systems).
As another poster rightly pointed out we should worry too much about a practical limit of 16 gigs, since there's not really a market demand for this in IA32. When it gets to be an issue somebody will find a way around it.
Of the people who might want >16 gigs of RAM right now I expect almost all of them are working with large databases of one form or another and could work around the hassles of 4m pages, thus scaling
all the way to 64 gig.
While theoretically you can put up to 64 gigs into a recent IA32 machine, my understanding is that in Linux at least the practical limit is 16 gigabytes. After that the page tables won't fit in kernel space.
I'd expect that most other OSes have limits like that due to architectures designed when nobody was close to using a full 4 gigs.
Actually I don't know that the majority of Xfree86 users do not use remote X. I can't think of anyone I've ever known who uses X and doesn't use the remoting feature at least occasionally.
For my part, it's almost always the case that the only applications that are local are my web browser and my mail client.
If you are part of an organization that has a Unix infrastucture (such as a university or many computer industry businesses) and use Unix yourself, you are almost certainly a user of remote X.
I guess my point would be that if you have more than one Unix machine you probably use X remoting at least some of the time. Unless maybe you're from a Windows/Mac background and just aren't used to the convenience of X.
While we're at it I think Thompson owns RCA. RCA is these days just a brand of some European company in any case.
My laptop has a single dead pixel, or more accurately 1/3 of a dead pixel. It will show green and blue but not red. It's slightly annoying but you get used to it.
If you really want to stay within the bounds of science, the principle of parsimony only applies when you actually have a theory. "I don't know how the universe was created" isn't a theory, so you can't eliminate "God created the universe" on the grounds that the first is simpler.
By way of analogy, suppose I don't know how electricity. My statement to that effect is more parsimonious than your explanation that the phenomenon we observe as electricity is the flow of electrons, but that doesn't matter because in this case you have a theory and I don't.
It's also important to understand that science does not purport to have THE answer. It purports to have AN answer, and in particular a useful answer. The principles of science prefer certain kinds of answers to others (falsifiable, parsimonious, etc.)