Re:Question for Iridium-knowledgeable, haiku form
on
A Eulogy for Iridium
·
· Score: 2
No offense, but 2400 (I've also heard 9600) baud per channel is a basically irrelevant figure AFAICT. One could design a receiver to aggregate transmission and reception across multiple channels if really necessary (somewhat like multichannel Ethernet aka Ethernet NIC bonding.) The bandwidth per satellite to and from ground is the key question needing an answer to understand the overall economic wisdom of spending $650 million annually (~$10 mil per sat) to keep em in the air.
--LP
Question for Iridium-knowledgeable, haiku form
on
A Eulogy for Iridium
·
· Score: 2
"Birds are too costly -- six-fifty mil," they say. But what bandwidth per bird?
As I read several articles this morning, the MS concessions appeared to be allowing some access to some OS code to some OEMs and other software developers. This is not the same as a generic offer to open the source to any interested party. The criteria for who can view the source and under what licensing conditions is obviously a rather material issue.
Say for example that "serious" companies (serious defined as having a certain revenue level) were allowed to access the source. That'd still lockout startups and open source efforts. Not to mention the various ways in which the license that is written "to protect Microsoft's intellectual property" could in turn restrict the ability for people to write GPL-resulting code.
In short, the fine print is rather important. Do you trust Microsoft, whose legal accumen has been a critical weapon since its first DOS license agreement and before, to submit any settlement terms that wouldn't have several loopholes they can rush through?
I pored through the Java contract Sun wrote when it became public, and believe me, that one was pretty tight and there were still a few subtle but critical holes that MS was able to drive a Mac truck through.
The very art of compromise and give and take the DOJ would need to eliminate the big holes would still leave need small holes MS could exploit steadily and with substantial strategic success due to their market power.
I remain highly skeptical that any conduct remedies could succeed (in fulfilling the purposes of the Sherman Anti-trust Act) in the face of a relentless, aggressive, and highly intelligent competitor with immense market (monopoly) power.
Don't be stupid. Noone in their right mind wants to see Microsoft broken up. It's no good for anyone. It'd be nice if they opened up all their APIs, and curbed some of their business practices, but a splintered Microsoft is an awful remedy.
As a small long-term (5+ years) stockholder of Microsoft, I wouldn't mind the company being broken up rather than to have some restrictive remedies. In five more years, I'm confident that I'd still be doing better than the market index.
The remedies you suggest are pointless:
Opening APIs without opening source is irrelevant; knowing the syntax of calls without opening the precisely-defined semantics (bugs and all), something really only available with reference source code, is useless for the purpose of interoperability, which is what customers most want between their various software programs; this is one (of several) important reasons why Linux has succeeded as a centralized UNIX platform when all past UNIX standards efforts have failed.
The hard-charging culture and high intelligence at Microsoft (along with a dose of legal savvy that has been with the firm since the DOS license agreement) will basically make "curbing business practices" largely ineffective at solving the issue of Microsoft's wielding of monopoly power. In two years, the issue will just reassert itself, and we'll be back to square one. While Microsoft management seems to think that stalling tactics would be profitable (I'd agree this is factually true; every day cements their dominance further), such tactics are shortsighted evasions of a willingness to recognize the validity and purpose of anti-trust law, and the refusal to tackle the real problem of distrust and fear that their previous behavior and relationship with partners has earned them from friend and foe alike. (Try naming one serious Microsoft partnership that has succeeded. They're like the bully of the playground who wins and keeps trying to but in when people leave to play another game. A breakup would help break this cycle of counter-productive dominance.)
As a stockholder who has watched his small MS holdings go up 10x in the last five years (wish I'd bought 10x more!) I wouldn't mind a short-term dip in return for a more workable, less threatening long-term corporate structure.
Sure, as a stockholder I have some concern that a dip would cause employees to flee their golden handcuffs, but at the end of the day, this is like a gust of wind carrying a huge load of acorns off a heavily laden tree. Like a tree struggling to reach the sun (justifying itself by describing the increase in square footage of shade for consumers such an effort would generate,) Ballmer et all can't succeed in their struggle against the law of large numbers forever without getting smaller and more focused. Microsoft would be able to play more effectively in a lot more strategic markets (e.g. WinCE) if they presented less of a threat to everyone else. A breakup would help them reduce their threat profile more effectively than any API opening or business practice remedies that would leave them on their enjoyable but pointless treadmill.
The scientist is only supposed to build and test theories, but the engener actually needs to figure out what to use. The science provides the engeners with something concreate to argue over the relivance to a specific problem....
This stands in contrast to religion where there is no science side (or they are combined in the case of new religions and cults which can change their rules). The divide between science and engenering helps both stay more objective.. and religion lacks this divide....
Actually, some religions have a flavor of this natural-world-objectivity check&balance scheme. With Christianity, there is the principle that, besides examining scripture, you shall know (judge) a false teacher and false teaching by examining its fruits. A false teacher bears bad fruit, a good teacher bears good fruit. Philosophically, this proposition alone is obviously not airtight, ("define good fruit") but it does form a type of hedge against extremism which has clearly negative effects. And it's useful for Christians to keep in mind as they examine their own history.
Other Christian appeals to real-world objectivity might include the passage "you shall know the truth, and the truth shall make you free."
You might offer Coke's lawyers that in return for dropping the suit, you'd be glad to sign an agreement binding yourself not to associate the site in any way with Coca-cola or any other beverage products.
[Or if that doesn't satisfy them, agree to only setup the site with material specifically oriented around cocaine, and/or setup tighter clauses about who you can sell the domain name to so they can't be burned that way.]
Hiring a lawyer to do this negotiation wouldn't be a bad idea if you're serious; otherwise, they'll probably sue and you'll end up as little guy bug meat, hate to say it. Unless you can get the EFF or someone on your side.
You didn't say whether you'd actually put up any material on the website dealing with cocaine; obviously actions of that sort would strengthen your argument and case, as would marketing/promoting your site as a source of knowledge about cocaine. It'd be much harder for Coke to convince a judge you are infringing when you have spent time and effort (and money) on meeting those goals that your motive is trademark infringing.
If you only break laws when there is a strong and compelling reason the law is wrong, one could argue that is a form of ethical civil disobedience. However, keep in mind that disobeying *any* law you think is wrong (and encouraging everone in society to do the same) essentially yields anarchy (no rule of law) since each person can claim their own ethical code, right?
I haven't objected to the ethicality of Mr. Haselton's actions; I was merely questioning the legality which looked pretty shaky to me. If we want the laws to be changed or bent or interpreted in our favor, we have to get better test cases than this one going through the courts, at least in my view.
This is a nice argument (function = restricting flow of info = security, audits as fundamental to security.) It's less torturous than the ones I was considering. I'll have to think about it some more about the redefinitions it implies to consider whether they really could be plausible and fairly applied in a neutral court setting. Intuitively I'm skeptical, but it's a stronger argument than those I've considered so far.
I agree that the DMCA is a bad law for consumers.
--LP
(Moderators: moderate up the parent post please?;-)
I think a consumer has a RIGHT to do whatever they wish to a product that they purchase.
I agree with you that we *should* have those rights. Legally, we don't though.
I'd point out a minor expansion of your comment; we don't just need those rights for product we purchase, we need those rights for any product we license a right to use (since that is how most software is "sold"-- as a license to use rather than an outright purchase.) Again, IANAL, but these rights would have to be strong enough to override normal rights of two parties to enter into a contract. It will take a substantial excercise of political power to get such rights passed in the face of entrenched corporate interest.
My point is that testing-the-software-to-ensure-it-works-properly (i.e. blocks sites properly) is not the same as "reverse engineering to test security methods" or "reverse engineering to insure interoperability between two programs." I agree that it would be valuable to be able to reverse engineer programs to make sure they "work properly," but that is not one of the fair use protections described in the posts I've read that worry about the DMCA.
Neither the DeCSS guys nor Mr Haselton appear to have given careful thought as to how to avoid prosecution under statutes like DMCA; if Mr Haselton had taken more care to avoid posting decrypted contents and had started out designing a third-party software package that would require decrypting iGear's list, he might have had a much better legal defense, but right now, it looks pretty weak to me given the existing language of the law.
(Of course, I agree that the existing law should be changed. I don't like DMCA either. But it still looks pretty cut-n-dry to me that Mr Haselton broke it. I guess time will tell.)
Thanks for the clarification. I should have said "once encrypted contents". My point still stands however. Mr Haselton's publication of the once-encrypted contents along with an analysis of the contents, (not just publishing tools or an alert that the security was weak) suggests quite strongly that his goal was *not* testing security methods but gaining access to secured content. If he had just published the code, he'd have a much stronger argument. The actual number of URLs posted and analyzed is fairly irrelevant. Whether you publish 50 of the URLs or all of them, you have still posted some of the once-encrypted contents, and if the DMCA applies, Mr. Haselton is in legal trouble AFAICT.
Blind assertions don't make truth or a good legal defense.
How does Haselton's cracking honestly fall under the definition of "interoperability" or "testing computer security systems"? Any definition I can think of where Haselton's actions would be considered "testing security" would be so tortuous as to render the phrase meaningless. "No sir, I wasn't hacking the encryption, I was just testing security systems" isn't going to fly without additional credible indication of intent. Mr Haselton's publication of the encrypted contents along with an analysis of the contents, (not just publishing the fact that the security was weak like 99% of security alerts) suggests quite strongly that his goal was *not* testing security methods but gaining access to secured content. The interoperability argument in this case is even more specious-- what two pieces of software was Mr. Haselton trying to make interoperate?
IANAL, but Haselton looks like he's standing on shaky ground, even assuming a noble purpose. Looks to me like a classic case of thinking that the ends justify the means. I welcome rational counterarguments; perhaps I'm missing something?
This is fundamentally false. Processor prices and memory prices drop at essentially the same rate since they are based on similar manufacturing technology. There are bubbles over time in pricing if Intel gets lazy or agressive or if the RAM manufacturers don't build enough factories or conspire to keep prices high, but overall, for the forseeable future both microprocessors and RAM will follow Moore's Law.
It is true however that CPU performance improves much faster than *memory latency*. For this reason, for about the last decade, academics have speculated and examined the possibility that (simplistically speaking) CPUs might be built around RAM, rather than RAM built around CPUs.
To some extent, this thinking and today's reality match the scenario you outline; on-chip and off-chip RAM caches are taking up a steadily increasing percentage of chip real estate; for some chips, 3/4ths of the processor is transistors and paths for the cache memory. For economic reasons however, it will continue to make sense for quite some time to have full system memory implemented separately from the CPUs.
I recognize this as way too late to get moderated up but perhaps at least you will notice this extra reply.
When this technology was first publicly disclosed a year ago or so, I talked to several people at C3D and questioned them about the 1 giga*byte* rates described on their home pages. (Search past threads for my handle and C3D). Since this bandwidth was as fast as RAM and an even bigger breakthrough than the capacity jump I was intrigued and highly skeptical.
You're right that they get one factor of 10-100x from the extra layers. Disk bandwidth grows essentially linearly with the number of layers. But the other factor of 10-100x comes from another technique (and there's a catch.) They can split the laser beam into multiple beams and read the results with a CCD-like sensor array for parallel read capability. Cool. This is described somewhat in the white paper on their website, if I recall correctly. The big catch is that this parallel-read capability does not occur with the high-capacity round spinning disks, but only occurs with the smaller flash-card form factor devices that don't rotate (and have simple rectangular regions for data, I suspect).
Unfortunately this suggests a technology with a very small market niche. For starters, most I/O busses actually run at slower speeds, creating somewhat of a disconnect until Intel's PCI and PCI-X successor "Infiniband" comes out in 2002 or whenever... And what good are 1 GByte/sec read rates when they're for a flash-card device that only holds 10 GB?
--LP
P.S. Read-write was also a serious technical challenge requiring significantly different materials when I talked to them ~9 months ago. Keep your eye out for progress on that.
You didn't define what you meant by "entering the second Internet era", but I'd say we're entering the third Internet era, not the second. The three eras I see would be:
First era, "Internet for techies", 1969-1993: Internet technology develops, rapidly grows in size as educational network and tech-corporate email gateway, spurred by government research grants and applications like email and USENET
Second era, "Internet for the masses", 1994-today: Internet enters into widespread use by consumers and businesses, spurred by development of the web browser graphical interface
Third era, "Broadband Internet", 1999-tomorrow: cable modem and DSL infrastructures remove bandwidth constraints and enable mass-market content delivery of all media types, spurred by the development of erbium-doped fiber optical amplifiers and dense wave division multiplexing.
If you actually read even the articles there summarizing the findings, you'd see that the author addresses your points. To wit:
1D and 2D basically aren't complex enough to have life and self-aware creatures that could observe the universe (i.e. why "we" couldn't be in 2D universe)
you can model two or more time dimensions via mathematics. 2 time dimensions makes motion unpredictable, unpredicatable; self-awareness requires some ability to predict from "past" results
On point three, you don't seem to get it. You can't put fake info in their IPv6 packets without detection (and discard) being possible at each router in the network thanks to the authentication header (which acts like a digital signature.) IP spoofing can't be detected at the packet level unless you can make sufficient assumptions about the ever-changing network and program them into each of your routers.
And back to point two, tracking compromised systems is a huge benefit since it A) speeds up the time to shut down/notify offending sites *much* more rapidly, even if they were hacked, and B) makes things much riskier for the hackers attempting to carry out such attacks.
One would never do this with "every router"; at most, one would do this with routers on the "edge" of your network.
Even then, you're imposing a burden on routers and more importantly router administrators to configure each router appropriately. And (somewhat like IPv6 adoption), you are requiring everyone on the Internet to adopt a proceedure and process to make up for flawed technology. I'd call that a fundamental flaw.
Short-term, your tools help act as "virus-checker" type solutions. In terms of long-term solutions for DoS+spoofing attacks, the main one I've seen proposed is to convince all ISPs to filter their outbound traffic to prevent outbound spoofing of packets claiming to come from other networks.
Given that IP spoofing is a fundamental flaw in IPv4, does this rise of spoofing-abetted DoS attacks increase the potential value of moving networks to IPv6 (with its per-packet authentication headers)? What solution would be best from your point of view?
How did they determine that particular split (1.35M servers, 3.9M clients) between Linux copies being sold for servers and ones being sold for workstations? This is easy with NT and with the UNIXes given the branding and packaging issues there, but totally non-trivial for Linux.
Some major assumption is buried here. I'd like to know what it is and whether it has held constant. It could affect that growth rate (98%) significantly.
Other than that, glad to see continued progress according to this metric. Useful for talking to those PHBs.:-)
Slashdot Mirror
No offense, but 2400 (I've also heard 9600)
baud per channel is a basically irrelevant
figure AFAICT. One could design a receiver
to aggregate transmission and reception across
multiple channels if really necessary (somewhat
like multichannel Ethernet aka Ethernet NIC bonding.) The bandwidth per satellite to and from ground is the key question needing an answer to understand the overall economic wisdom of spending $650 million annually (~$10 mil per sat) to keep em in the air.
--LP
"Birds are too costly --
six-fifty mil," they say. But
what bandwidth per bird?
Say for example that "serious" companies (serious defined as having a certain revenue level) were allowed to access the source. That'd still lockout startups and open source efforts. Not to mention the various ways in which the license that is written "to protect Microsoft's intellectual property" could in turn restrict the ability for people to write GPL-resulting code.
In short, the fine print is rather important. Do you trust Microsoft, whose legal accumen has been a critical weapon since its first DOS license agreement and before, to submit any settlement terms that wouldn't have several loopholes they can rush through?
I pored through the Java contract Sun wrote when it became public, and believe me, that one was pretty tight and there were still a few subtle but critical holes that MS was able to drive a Mac truck through.
The very art of compromise and give and take the DOJ would need to eliminate the big holes would still leave need small holes MS could exploit steadily and with substantial strategic success due to their market power.
I remain highly skeptical that any conduct remedies could succeed (in fulfilling the purposes of the Sherman Anti-trust Act) in the face of a relentless, aggressive, and highly intelligent competitor with immense market (monopoly) power.
--LP
As a small long-term (5+ years) stockholder of Microsoft, I wouldn't mind the company being broken up rather than to have some restrictive remedies. In five more years, I'm confident that I'd still be doing better than the market index.
The remedies you suggest are pointless:
Opening APIs without opening source is irrelevant; knowing the syntax of calls without opening the precisely-defined semantics (bugs and all), something really only available with reference source code, is useless for the purpose of interoperability, which is what customers most want between their various software programs; this is one (of several) important reasons why Linux has succeeded as a centralized UNIX platform when all past UNIX standards efforts have failed.
The hard-charging culture and high intelligence at Microsoft (along with a dose of legal savvy that has been with the firm since the DOS license agreement) will basically make "curbing business practices" largely ineffective at solving the issue of Microsoft's wielding of monopoly power. In two years, the issue will just reassert itself, and we'll be back to square one. While Microsoft management seems to think that stalling tactics would be profitable (I'd agree this is factually true; every day cements their dominance further), such tactics are shortsighted evasions of a willingness to recognize the validity and purpose of anti-trust law, and the refusal to tackle the real problem of distrust and fear that their previous behavior and relationship with partners has earned them from friend and foe alike. (Try naming one serious Microsoft partnership that has succeeded. They're like the bully of the playground who wins and keeps trying to but in when people leave to play another game. A breakup would help break this cycle of counter-productive dominance.)
As a stockholder who has watched his small MS holdings go up 10x in the last five years (wish I'd bought 10x more!) I wouldn't mind a short-term dip in return for a more workable, less threatening long-term corporate structure.
Sure, as a stockholder I have some concern that a dip would cause employees to flee their golden handcuffs, but at the end of the day, this is like a gust of wind carrying a huge load of acorns off a heavily laden tree. Like a tree struggling to reach the sun (justifying itself by describing the increase in square footage of shade for consumers such an effort would generate,) Ballmer et all can't succeed in their struggle against the law of large numbers forever without getting smaller and more focused. Microsoft would be able to play more effectively in a lot more strategic markets (e.g. WinCE) if they presented less of a threat to everyone else. A breakup would help them reduce their threat profile more effectively than any API opening or business practice remedies that would leave them on their enjoyable but pointless treadmill.
--LP
This stands in contrast to religion where there is no science side (or they are combined in the case of new religions and cults which can change their rules). The divide between science and engenering helps both stay more objective.. and religion lacks this divide. ...
Actually, some religions have a flavor of this natural-world-objectivity check&balance scheme. With Christianity, there is the principle that, besides examining scripture, you shall know (judge) a false teacher and false teaching by examining its fruits. A false teacher bears bad fruit, a good teacher bears good fruit. Philosophically, this proposition alone is obviously not airtight, ("define good fruit") but it does form a type of hedge against extremism which has clearly negative effects. And it's useful for Christians to keep in mind as they examine their own history.
Other Christian appeals to real-world objectivity might include the passage "you shall know the truth, and the truth shall make you free."
--LP
Hmm, gravity too
weak for coalescence to
occur frequently?
Interesting. Thanks.
Runty planets--
even brown dwarfs have more mass!
(and get warm at night...)
Vagabond planets--
if enough collide, do they
alight with fusion?
I'm fairly sure that Tru64 UNIX's AdvFS (from Compaq/Digital) provides such snapshotting.
--LP
You might offer Coke's lawyers that in return for dropping the suit, you'd be glad to sign an agreement binding yourself not to associate the site in any way with Coca-cola or any other beverage products.
[Or if that doesn't satisfy them, agree to only setup the site with material specifically oriented around cocaine, and/or setup tighter clauses about who you can sell the domain name to so they can't be burned that way.]
Hiring a lawyer to do this negotiation wouldn't be a bad idea if you're serious; otherwise, they'll probably sue and you'll end up as little guy bug meat, hate to say it. Unless you can get the EFF or someone on your side.
You didn't say whether you'd actually put up any material on the website dealing with cocaine; obviously actions of that sort would strengthen your argument and case, as would marketing/promoting your site as a source of knowledge about cocaine. It'd be much harder for Coke to convince a judge you are infringing when you have spent time and effort (and money) on meeting those goals that your motive is trademark infringing.
--Greg
I agree totally.
--LP
If you only break laws when there is a strong and compelling reason the law is wrong, one could argue that is a form of ethical civil disobedience. However, keep in mind that disobeying *any* law you think is wrong (and encouraging everone in society to do the same) essentially yields anarchy (no rule of law) since each person can claim their own ethical code, right?
I haven't objected to the ethicality of Mr. Haselton's actions; I was merely questioning the legality which looked pretty shaky to me. If we want the laws to be changed or bent or interpreted in our favor, we have to get better test cases than this one going through the courts, at least in my view.
--LP
This is a nice argument (function = restricting flow of info = security, audits as fundamental to security.) It's less torturous than the ones I was considering. I'll have to think about it some more about the redefinitions it implies to consider whether they really could be plausible and fairly applied in a neutral court setting. Intuitively I'm skeptical, but it's a stronger argument than those I've considered so far.
;-)
I agree that the DMCA is a bad law for consumers.
--LP
(Moderators: moderate up the parent post please?
I think a consumer has a RIGHT to do whatever they wish to a product that they purchase.
I agree with you that we *should* have those rights. Legally, we don't though.
I'd point out a minor expansion of your comment; we don't just need those rights for product we purchase, we need those rights for any product we license a right to use (since that is how most software is "sold"-- as a license to use rather than an outright purchase.) Again, IANAL, but these rights would have to be strong enough to override normal rights of two parties to enter into a contract. It will take a substantial excercise of political power to get such rights passed in the face of entrenched corporate interest.
--LP
My point is that testing-the-software-to-ensure-it-works-properly (i.e. blocks sites properly) is not the same as "reverse engineering to test security methods" or "reverse engineering to insure interoperability between two programs." I agree that it would be valuable to be able to reverse engineer programs to make sure they "work properly," but that is not one of the fair use protections described in the posts I've read that worry about the DMCA.
Neither the DeCSS guys nor Mr Haselton appear to have given careful thought as to how to avoid prosecution under statutes like DMCA; if Mr Haselton had taken more care to avoid posting decrypted contents and had started out designing a third-party software package that would require decrypting iGear's list, he might have had a much better legal defense, but right now, it looks pretty weak to me given the existing language of the law.
(Of course, I agree that the existing law should be changed. I don't like DMCA either. But it still looks pretty cut-n-dry to me that Mr Haselton broke it. I guess time will tell.)
--LP
Thanks for the clarification. I should have said "once encrypted contents". My point still stands however. Mr Haselton's publication of the once-encrypted contents along with an analysis of the contents, (not just publishing tools or an alert that the security was weak) suggests quite strongly that his goal was *not* testing security methods but gaining access to secured content. If he had just published the code, he'd have a much stronger argument. The actual number of URLs posted and analyzed is fairly irrelevant. Whether you publish 50 of the URLs or all of them, you have still posted some of the once-encrypted contents, and if the DMCA applies, Mr. Haselton is in legal trouble AFAICT.
--LP
How does Haselton's cracking honestly fall under the definition of "interoperability" or "testing computer security systems"? Any definition I can think of where Haselton's actions would be considered "testing security" would be so tortuous as to render the phrase meaningless. "No sir, I wasn't hacking the encryption, I was just testing security systems" isn't going to fly without additional credible indication of intent. Mr Haselton's publication of the encrypted contents along with an analysis of the contents, (not just publishing the fact that the security was weak like 99% of security alerts) suggests quite strongly that his goal was *not* testing security methods but gaining access to secured content. The interoperability argument in this case is even more specious-- what two pieces of software was Mr. Haselton trying to make interoperate?
IANAL, but Haselton looks like he's standing on shaky ground, even assuming a noble purpose. Looks to me like a classic case of thinking that the ends justify the means. I welcome rational counterarguments; perhaps I'm missing something?
--LP
This is fundamentally false. Processor prices and memory prices drop at essentially the same rate since they are based on similar manufacturing technology. There are bubbles over time in pricing if Intel gets lazy or agressive or if the RAM manufacturers don't build enough factories or conspire to keep prices high, but overall, for the forseeable future both microprocessors and RAM will follow Moore's Law.
It is true however that CPU performance improves much faster than *memory latency*. For this reason, for about the last decade, academics have speculated and examined the possibility that (simplistically speaking) CPUs might be built around RAM, rather than RAM built around CPUs.
To some extent, this thinking and today's reality match the scenario you outline; on-chip and off-chip RAM caches are taking up a steadily increasing percentage of chip real estate; for some chips, 3/4ths of the processor is transistors and paths for the cache memory. For economic reasons however, it will continue to make sense for quite some time to have full system memory implemented separately from the CPUs.
--LP
When this technology was first publicly disclosed a year ago or so, I talked to several people at C3D and questioned them about the 1 giga*byte* rates described on their home pages. (Search past threads for my handle and C3D). Since this bandwidth was as fast as RAM and an even bigger breakthrough than the capacity jump I was intrigued and highly skeptical.
You're right that they get one factor of 10-100x from the extra layers. Disk bandwidth grows essentially linearly with the number of layers. But the other factor of 10-100x comes from another technique (and there's a catch.) They can split the laser beam into multiple beams and read the results with a CCD-like sensor array for parallel read capability. Cool. This is described somewhat in the white paper on their website, if I recall correctly. The big catch is that this parallel-read capability does not occur with the high-capacity round spinning disks, but only occurs with the smaller flash-card form factor devices that don't rotate (and have simple rectangular regions for data, I suspect).
Unfortunately this suggests a technology with a very small market niche. For starters, most I/O busses actually run at slower speeds, creating somewhat of a disconnect until Intel's PCI and PCI-X successor "Infiniband" comes out in 2002 or whenever... And what good are 1 GByte/sec read rates when they're for a flash-card device that only holds 10 GB?
--LP
P.S. Read-write was also a serious technical challenge requiring significantly different materials when I talked to them ~9 months ago. Keep your eye out for progress on that.
First era, "Internet for techies", 1969-1993: Internet technology develops, rapidly grows in size as educational network and tech-corporate email gateway, spurred by government research grants and applications like email and USENET
Second era, "Internet for the masses", 1994-today: Internet enters into widespread use by consumers and businesses, spurred by development of the web browser graphical interface
Third era, "Broadband Internet", 1999-tomorrow: cable modem and DSL infrastructures remove bandwidth constraints and enable mass-market content delivery of all media types, spurred by the development of erbium-doped fiber optical amplifiers and dense wave division multiplexing.
--LP
1D and 2D basically aren't complex enough to have life and self-aware creatures that could observe the universe (i.e. why "we" couldn't be in 2D universe)
you can model two or more time dimensions via mathematics. 2 time dimensions makes motion unpredictable, unpredicatable; self-awareness requires some ability to predict from "past" results
Not bulletproof of course, but thought-provoking.
--LP
On point three, you don't seem to get it. You can't put fake info in their IPv6 packets without detection (and discard) being possible at each router in the network thanks to the authentication header (which acts like a digital signature.) IP spoofing can't be detected at the packet level unless you can make sufficient assumptions about the ever-changing network and program them into each of your routers.
And back to point two, tracking compromised systems is a huge benefit since it A) speeds up the time to shut down/notify offending sites *much* more rapidly, even if they were hacked, and B) makes things much riskier for the hackers attempting to carry out such attacks.
--LinuxParanoid
One would never do this with "every router"; at most, one would do this with routers on the "edge" of your network.
Even then, you're imposing a burden on routers and more importantly router administrators to configure each router appropriately. And (somewhat like IPv6 adoption), you are requiring everyone on the Internet to adopt a proceedure and process to make up for flawed technology. I'd call that a fundamental flaw.
--LP
Short-term, your tools help act as "virus-checker" type solutions. In terms of long-term solutions for DoS+spoofing attacks, the main one I've seen proposed is to convince all ISPs to filter their outbound traffic to prevent outbound spoofing of packets claiming to come from other networks.
Given that IP spoofing is a fundamental flaw in IPv4, does this rise of spoofing-abetted DoS attacks increase the potential value of moving networks to IPv6 (with its per-packet authentication headers)? What solution would be best from your point of view?
--LP
How did they determine that particular split (1.35M servers, 3.9M clients) between Linux copies being sold for servers and ones being sold for workstations? This is easy with NT and with the UNIXes given the branding and packaging issues there, but totally non-trivial for Linux.
:-)
Some major assumption is buried here. I'd like to know what it is and whether it has held constant. It could affect that growth rate (98%) significantly.
Other than that, glad to see continued progress according to this metric. Useful for talking to those PHBs.
--LP