Probably the one that most thoroughly examined the inherent tradeoffs, and that's most directly based on Shamir's IDA work, is PASIS at CMU.
PASIS is not based on IDA. PASIS is not a mechanism for data dispersal. It is a family of storage protocols that make efficient use of data dispersal mechanisms. Any mechanism that satisfies the m-of-n condition (where of n data fragments, m are necessary to reconstruct the orignal data item) can be used. This can be mirroring, striping, erasure coding, IDA, and anything else researchers will cook up in the future.
PASIS extends combines data dispersal with checksumming to not only deal with lost data, but bad data. Additionally, it can be tuned to address different fault models, such as fail-stop (where machines that go down never come back up), or synchronous timing (where working machines are guaranteed to respond within a predefined timeout period).
The motivation behind PASIS is to expose all options and possibilties. Few questions are actually answered. It shows you all the knobs and switches, but you still have to dial in the right setting. Any practical implementation of a distributed storage system (RAID, Google FS, Cleversafe, etc) has to answer these questions.
Threads are used because they are easy for development. They can also keep the multiple units of a parallel processor busy. However, using more threads than there are processing units introduces overhead into the system. There are better ways to do task scheduling... like event-driven models.
So, why not combine the multi-threaded and event-driven models? Some very interesting research has been done in this area. Check out Staged Event-Driven Architecture, or SEDA. Like threads, it has high fairness in scheduling. Like event-driven models, it scales to a large number of concurrent requests. In fact, it degrades gracefully even during a Slashdot effect.
It's a shame that WinFS has been dropped. However, Vista has TxF, which is a framework that allows for database-like ACID transactional semantics in an NTFS file system. This is an awesome advancement in filesystems, and a similar model should be proposed for the POSIX standard.
Last time I checked, there exists a project already doing this. It's called SETI. They very thouroughly comb a large range of the EM spectrum for any data representing intelligent life. This proposal instead takes hazy samples from a very narrow band of data (the visible spectrum), to guess at the chemical composition of other planets. So we've replaced listening to white noise with looking at faint blue dots.
One of the problems with Mr. Fusion is that it would produce way too much energy. One banana peel, into pure energy, would produce 1.25 billion kilowatt-hours. How many miles can you get on that? Releasing such energy instantaneously would probably spell the end of this sector of the solar system.
I feel they have no reason to complain. People searching for pirated software generally know how to find it, and don't need Google Suggest. People not searching to pirate software won't suddenly get that desire because it was suggested by Google Suggest. In fact, ServersCheck could use this to their advantage. By using Google Suggest, it is now easier for them to find the bad guys that are pirating their software.
In about 10 years you're going to be able to buy single wireless access points from Best Buy that will cover the size of the city and it's bandwidth needs for about 50 USD.
Not going to happen given how the FCC manages spectrum and transmission power.
Good point. Technology changes rapidly, not the laws of physics. For an access point to have that kind of capability, it would dump pretty high-power transmissions into the EM spectrum. While it is possible that the technology to do that will improve, by becoming smaller and cheaper, the issue of regulation still remains. It would likely cause tons of interference for those near the transmission point, and would possibly even pose a health hazard. Notice that in the evolution of wireless technology, bandwidth improves at a much higher rate than transmission range.
Also, it's a fact that wireless broadcast technology isn't going to look like that. By that logic, television broadcast towers should soon be the size of a car antenna, and cell phone towers should soon be the size of a fingernail. These technologies don't seem to be moving in that direction.
Why is he spreading FUD against Boot Camp? Half of his points are irrelevant, and the other half are flat out false.
So what happens when that critical new part, new application, or new Web site fails to run? Who is going to help? Apple? Hardly.
As if someone was there to help with my PC. Who am I going to call when my Web site fails to run... definitely not the manufacturer that made the machine. And most hardware in the Mac is going to be equivalent to PC hardware, so why should I have a problem servicing it?
Apple's not interested in a DIY Mac, nor is it concerned with the case-mod culture of the PC.
Big deal.If someone is smart enough to case-mod ther PC... (1) they will be able to case-mod their Mac... why would this even be a problem? Is there a force field blocking people from going inside their cases? (2) they will probably be able to get Mac OS X on a PC
Apple's not really known as the low-cost or value leader when it comes to computers. There's a premium to pay for "Think Different" that has nothing to do with performance or capability.
It's the consumer's choice to decide whether or not Steve Jobs is ripping them off. If it's worth it, then they will spend the money. What else is there to say? Why insult them for disagreeing with you?
Yesterday, I installed the Google Video Player standalone application on my system. When it finished, the installer presented me with a checkbox for setting Google my default search engine. It was checked by default, and I was about to blindly hit "Finish" without giving the widget any thought. Most people would probably do the same.
Usually, the last dialog in an installation asks the user about adding items to the start menu, the desktop, or other such means of starting the application. How does Google Video have any relation to Google Search? Why should the Google Video installer try to hook into my browser and set up Google Search? I don't know.
It's pretty apparent that both companies highly value having their search engines as users' defaults. And I think both companies are trying to accomplish this however they can. Google is complaining because they have a good image in the public's eye, and hope to leverage Microsoft's image as a monopoly to generate some public outrage in their favor. There's no right or wrong here, it's just one company versus another.
If I assign a value to a variable... great! I will do anything I want with the variable! Usage of a function pointer variable should not generate warnings. I'm talking about dubious usage of the function. If my code said...
(geteuid != 0)
... a warning should be generated. If my code said...
geteuid_var = geteuid; (geteuid_var != 0)
... then a warning should definitely not be generated. The compiler can differentiate between function and function pointer variable by looking at their respective declarations.
I would expect that something like this would through a warning, but it doesn't. I just tried this in GCC, and no warning occurred, not even with -Wall.
I would change the C standard to throw a warning on dubious use of function addresses. Practically, they are only used as part of an rval of an assignment operation, with the left hand side being a function pointer variable. If they are used in any other way, it's probably a typo.
Name me another way function addresses are commonly used.
AFAIK, this technology would only address a number of physical security threats. Adversaries would be stopped from stealing hard drives, or trying to pick up any bits which leak into the environment (maybe through EM signals). While these problems are important to solve, this technology is far from a be-all end-all solution.
Since the protection only occurs in hardware, one can still exploit the same software-based attack vectors that have been around for ages. Encryption is done even below the OS. If some Trojan horse got kernel-level priveleges and looked at physical RAM, it would see the plaintext and not the ciphertext, All the problems of network security still abound. as data sent out on the network is not protected.
I'm not sure how this could be used for DRM. I guess Secure Blue could be extended to give or take keys over the network, and data can be transmitted as ciphertext and stored as-is into RAM. Can someone explain how this would work? TPMs can be used for DRM, but they work quite differently.
I disagree that the screen is the only place for seeing data unencrypted. Devices will need DMA access to memory, and Secure Blue would have to decrypt the data before sending it down the bus. How is leakage protected in this case?
Like the above posters said, water is readily available and non-toxic, but the high specific heat capacity certainly does help. I think the only liquid with a higher specific heat capacity is ammonia.
Can someone explain to me why it's important that BMW is on Google anyway?
Isn't it a no-brainer that www.bmw.com is the global BMW site, and www.bmw.de is the German BMW site? Do people really need a search engine to find this out?
Soon, people are going to entirely replace DNS with Google.
I'm somewhat confused. Is Web 2.0 the movement towards building interactive applications on the Web (Ajax, Web services)... or is it gathering information in a more collaborative fashion (wikis, blogs, etc)?
I find people define Web 2.0 in both of these ways, and I don't understand how they are necessarily related. It seems to me that Web 2.0 is just a blanket term that covers however things are done now, Web 1.5 is however things were done before the bubble burst, and Web 1.0 is however things were done in the very beginning. I don't see cohesive paradigms that we can assign to the web as a whole; it seems to be more of a collection of general trends and technologies.
The Sega Saturn was the first video game console with multiple CPUs. One of the reasons why it flopped was that video game developers could not use both CPUs concurrently. They were stuck in a single threaded frame of mind.
Well, that was 10 years ago. I hope video game folks figured out multithreaded programming by now.
We can look at the transition in another area. Let's look at where the Linux kernel was 10 years ago. Linux 2.0 provided multiprocessing support, but only one CPU could be executing in kernel mode at a time (the big kernel lock). This kind of policy was chosen because it allowed the kernel to essentially remain a singlethreaded application. With each new release came incrementally more finely-grained locking. Developers were forced to rewrite their components to comply with the new standards and fix all the potential race conditions and deadlocks. By now, the Linux kernel can not only execute on multiple processors concurrently, it is preemptible too, allowing for concurrent round-robin execution of kernel threads on a single processor.
I understand that video games != OS kernels, but a similar transition could be made in the video game industry. It seems that the only obstacle is the complexity of multithreaded programming. Video games have great potential for being multithreaded, however, the staggering difficulty of it is readily apparent to me. Consider rendering a 3D space using multiple threads. Ouch.
Slashdot Mirror
PASIS extends combines data dispersal with checksumming to not only deal with lost data, but bad data. Additionally, it can be tuned to address different fault models, such as fail-stop (where machines that go down never come back up), or synchronous timing (where working machines are guaranteed to respond within a predefined timeout period).
The motivation behind PASIS is to expose all options and possibilties. Few questions are actually answered. It shows you all the knobs and switches, but you still have to dial in the right setting. Any practical implementation of a distributed storage system (RAID, Google FS, Cleversafe, etc) has to answer these questions.
Threads are used because they are easy for development. They can also keep the multiple units of a parallel processor busy. However, using more threads than there are processing units introduces overhead into the system. There are better ways to do task scheduling... like event-driven models.
So, why not combine the multi-threaded and event-driven models? Some very interesting research has been done in this area. Check out Staged Event-Driven Architecture, or SEDA. Like threads, it has high fairness in scheduling. Like event-driven models, it scales to a large number of concurrent requests. In fact, it degrades gracefully even during a Slashdot effect.
It's a shame that WinFS has been dropped. However, Vista has TxF, which is a framework that allows for database-like ACID transactional semantics in an NTFS file system. This is an awesome advancement in filesystems, and a similar model should be proposed for the POSIX standard.
Last time I checked, there exists a project already doing this. It's called SETI. They very thouroughly comb a large range of the EM spectrum for any data representing intelligent life. This proposal instead takes hazy samples from a very narrow band of data (the visible spectrum), to guess at the chemical composition of other planets. So we've replaced listening to white noise with looking at faint blue dots.
One of the problems with Mr. Fusion is that it would produce way too much energy. One banana peel, into pure energy, would produce 1.25 billion kilowatt-hours. How many miles can you get on that? Releasing such energy instantaneously would probably spell the end of this sector of the solar system.
That has to be updated. Google doesn't look like that anymore.
I feel they have no reason to complain. People searching for pirated software generally know how to find it, and don't need Google Suggest. People not searching to pirate software won't suddenly get that desire because it was suggested by Google Suggest. In fact, ServersCheck could use this to their advantage. By using Google Suggest, it is now easier for them to find the bad guys that are pirating their software.
Also, it's a fact that wireless broadcast technology isn't going to look like that. By that logic, television broadcast towers should soon be the size of a car antenna, and cell phone towers should soon be the size of a fingernail. These technologies don't seem to be moving in that direction.
Isn't child porn illegal? If so, then Google shouldn't be at fault, but the pornographers.
As if someone was there to help with my PC. Who am I going to call when my Web site fails to run... definitely not the manufacturer that made the machine. And most hardware in the Mac is going to be equivalent to PC hardware, so why should I have a problem servicing it?
Big deal.If someone is smart enough to case-mod ther PC...
(1) they will be able to case-mod their Mac... why would this even be a problem? Is there a force field blocking people from going inside their cases?
(2) they will probably be able to get Mac OS X on a PC
It's the consumer's choice to decide whether or not Steve Jobs is ripping them off. If it's worth it, then they will spend the money. What else is there to say? Why insult them for disagreeing with you?
So then, implanting an RFID tag into my breasts would probably not be a good idea...
Yesterday, I installed the Google Video Player standalone application on my system. When it finished, the installer presented me with a checkbox for setting Google my default search engine. It was checked by default, and I was about to blindly hit "Finish" without giving the widget any thought. Most people would probably do the same.
Usually, the last dialog in an installation asks the user about adding items to the start menu, the desktop, or other such means of starting the application. How does Google Video have any relation to Google Search? Why should the Google Video installer try to hook into my browser and set up Google Search? I don't know.
It's pretty apparent that both companies highly value having their search engines as users' defaults. And I think both companies are trying to accomplish this however they can. Google is complaining because they have a good image in the public's eye, and hope to leverage Microsoft's image as a monopoly to generate some public outrage in their favor. There's no right or wrong here, it's just one company versus another.
Ok, I agree. This is valid. Maybe then, there should be restrictions on implicit typecasting in expressions that contain function names.
If I assign a value to a variable... great! I will do anything I want with the variable! Usage of a function pointer variable should not generate warnings. I'm talking about dubious usage of the function. If my code said
Try a little bit harder next time.
I would expect that something like this would through a warning, but it doesn't. I just tried this in GCC, and no warning occurred, not even with -Wall.
I would change the C standard to throw a warning on dubious use of function addresses. Practically, they are only used as part of an rval of an assignment operation, with the left hand side being a function pointer variable. If they are used in any other way, it's probably a typo.
Name me another way function addresses are commonly used.
AFAIK, this technology would only address a number of physical security threats. Adversaries would be stopped from stealing hard drives, or trying to pick up any bits which leak into the environment (maybe through EM signals). While these problems are important to solve, this technology is far from a be-all end-all solution.
Since the protection only occurs in hardware, one can still exploit the same software-based attack vectors that have been around for ages. Encryption is done even below the OS. If some Trojan horse got kernel-level priveleges and looked at physical RAM, it would see the plaintext and not the ciphertext, All the problems of network security still abound. as data sent out on the network is not protected.
I'm not sure how this could be used for DRM. I guess Secure Blue could be extended to give or take keys over the network, and data can be transmitted as ciphertext and stored as-is into RAM. Can someone explain how this would work? TPMs can be used for DRM, but they work quite differently.
I disagree that the screen is the only place for seeing data unencrypted. Devices will need DMA access to memory, and Secure Blue would have to decrypt the data before sending it down the bus. How is leakage protected in this case?
Like the above posters said, water is readily available and non-toxic, but the high specific heat capacity certainly does help. I think the only liquid with a higher specific heat capacity is ammonia.
Get your suicide online, at www.eSeppuku.com
Can someone explain to me why it's important that BMW is on Google anyway? Isn't it a no-brainer that www.bmw.com is the global BMW site, and www.bmw.de is the German BMW site? Do people really need a search engine to find this out? Soon, people are going to entirely replace DNS with Google.
Why is it del.icio.us? Couldn't they just as easily have picked delicio.us?
I'm somewhat confused. Is Web 2.0 the movement towards building interactive applications on the Web (Ajax, Web services)... or is it gathering information in a more collaborative fashion (wikis, blogs, etc)?
I find people define Web 2.0 in both of these ways, and I don't understand how they are necessarily related. It seems to me that Web 2.0 is just a blanket term that covers however things are done now, Web 1.5 is however things were done before the bubble burst, and Web 1.0 is however things were done in the very beginning. I don't see cohesive paradigms that we can assign to the web as a whole; it seems to be more of a collection of general trends and technologies.
Nicely played. Although, your example does use JavaScript without tags, I think JavaScript in general needs to cleaned up.
HTML is fine. CSS is great. It's everything inside of that needs cleaning.
Yes, Google went to college!
It used to be google.stanford.edu. It was a part of Stanford's web site when it was still a research project.
The Sega Saturn was the first video game console with multiple CPUs. One of the reasons why it flopped was that video game developers could not use both CPUs concurrently. They were stuck in a single threaded frame of mind.
Well, that was 10 years ago. I hope video game folks figured out multithreaded programming by now.
We can look at the transition in another area. Let's look at where the Linux kernel was 10 years ago. Linux 2.0 provided multiprocessing support, but only one CPU could be executing in kernel mode at a time (the big kernel lock). This kind of policy was chosen because it allowed the kernel to essentially remain a singlethreaded application. With each new release came incrementally more finely-grained locking. Developers were forced to rewrite their components to comply with the new standards and fix all the potential race conditions and deadlocks. By now, the Linux kernel can not only execute on multiple processors concurrently, it is preemptible too, allowing for concurrent round-robin execution of kernel threads on a single processor.
I understand that video games != OS kernels, but a similar transition could be made in the video game industry. It seems that the only obstacle is the complexity of multithreaded programming. Video games have great potential for being multithreaded, however, the staggering difficulty of it is readily apparent to me. Consider rendering a 3D space using multiple threads. Ouch.