You obviously know little about recording. For starters, the analog sections of your PC's audio system are garbage by studio standards. And proper acoustically engineered space doesn't come cheaply either. Many studio-quality microphones (preamp not included) cost several times what the average PC costs.
Responding to your comment: Microsoft is the one prohibiting their licensees from benchmarking their database against others. They must have something to hide.
All commercial software vendors generally prohibit benchmarking to some extent these days. Yes, Oracle and IBM too. The idea is to prevent "bad" benchmarks by improperly configured setups. Most vendors' service organizations will assist journalists and even end-users with performing standard benchmarks (sometimes for a price).
As for SQL benchmarks, they do exist. And Microsoft SQL Server fares very well (espicially on a price/performance basis). See the non-clustered TPC-C benchmarks for four-socket servers for example. I've never seen a TPC result for an open source database listed. Why? Probably because MySQL and even PostGres can't scale like the commercial DBs, and nobody wants to invest time and money it takes to making a sucessful benchmark submission for a product they can't actually sell.
Note that Oracle has largely stopped submitting TPC results, as they've been getting their ass handed to them by MSFT and IBM so regularly.
Many web-based applications already do something better; my employer owns at least one.
The basic technique is to create a derived database access class(es) that include simple caching logic. The whole app then uses that derived class for DB access.
When the cache-aware class(es) are used, the developer indicates whether or not cached data is desiered, and optionally what the specific timeout should be. When a call to this "caching DB class" is made, the SQL that would be sent to the database is instead first hashed, and compared with the cache. If there is a match, and the cache timeout has not been exceeded, the data is returned directly from the cache, otherwise it is passed on to the DB and the result is cached.
Of course, the dev can always specify a query as not cacheable for transactional data, but common things like drop-downs and permissions look ups are quickly cached by this automatic mechanism. When our vendor implemented similar caching in their app, it reduced the average CPU utilization on our DB server by 40%, and made a noticeable improvement in end-user response times. And the amount of code they needed to change was very small; they only had to add 'cache=YES' to the DB access code in a very few critical areas of the application. But the key is that the whole app is still database driven: there are no static files to update when items like menus change.
Seriously, similar problems were predicted recently with the online streaming of "March Madness" (the national college basketball tournament) here in the USA. As a result, I got approval from Human Resources to pre-emptively block CBS Sports during tournament games. Apparently I wasn't the only network manager one who did this... many of my non-IT friends who work elsewhere called me asking "Hey, CBS is blocked, how can I get around this?"
The end result: networks didn't melt, and the Internet lived.
In any case, the streaming sources are limited for such events. The BBC (or other national media outlet) is likely to be overwhelmed before backbones or even end-user broadband connections become saturated.
For a start, libraries are public facilities (at least in my country; perhaps others differ) and subject to specific exemptions in some cases (same caveat applies). Google, OTOH, is a commercial organisation.
There are many private University and even non-academic private/corporate libraries to which the same fair use doctrine applies in the USA.
Google cache is really an archival snapshot of the web, much like the archive.org. It is fair use in that it reproduces what is publicly available in any case, but Google adds higher availability and searchability. The orgininal source is duly credited. It's much like a library saving the most recent issues of a newspaper, or the "news clipping" services that are often provided to Congress.
Anyway, I'm sure Google's lawyers understand the legal implications of their caching service far better than you and I, and yet they have not advised Google to take it down.
Traditional "fair use" doctrine, which protects the right of libararies to create things like card catalogs, or allows them to make microfilm copies for indexing and archival purposes. Google serves as a mostly indiscriminate "card catalog" for the Internet. The problem is, as soon as Google starts discriminating about what they index (other than by following robots.txt directions), they open themselves up to having to comply with a million different laws from a million different communites and nations. Google already started down this slippery slope with the Chinese.
MS, and a great number of others, embraced S/MIME over PGP for email encryption many years ago. And you know what? S/MIME just works in Outlook, and certificate management is about as easy as it can get with a PKI.
Now, why did Microsoft (and Netscape, Lotus, Novell, etc.) pick S/MIME over PGP?
Likely because the PGP web-of-trust model is impossible for non-technical users to understand. The WoT is still quite disconnected after all these years of PGP use. While in theory it scales infintely, in practice it doesn't work out so well. S/MIME works just like SSL, meaning the user doesn't have to worry too much about trust, the computer handles the PKI work.
Also, at the time S/MIME was integrated into Outlook, PGP was text-only, while S/MIME offered HTML and attachment support. Very few programs supported PGP/MIME in any reasonable fashion back in the late 90s, and from what I can see the majority of PGP email use still seems to be text-only to this day.
Not at all, older versions of the exchange server used to have separate mailboxes for users connecting via exchange and POP.
The above statement is totally and completely false. I have been administering Exchange servers since version 5.0 in the mid-90s, and there has only ever been one kind of mailbox. See my previous statements about the pssibilty that the people you reference may not have any understanding the archicture of Exchange Server.
This is, indeed, what we tried, but no one can seem to keep the server stable in this configuration.
Tens of thousands of geographically dispersed enterprises that run Exchange in mixed environments would disagree with you. I know admins at plenty of Exchange server shops besides mine, and none of them have ever experienced "instability" due to serving mail via POP and IMAP. By the way, if you could describe the nature of the instability, I would perhaps be more inclined to believe your reports. What happens, exactly?
"Exchange server pop problems" yields about 7 million hits in Google, so I don't think I'm alone.
Nice straw man. A Google for "linux pop problems" yields more than 13 million hits. By your logic, we must conclude that Linux can't be made into a functioning POP Server, no matter what POP daemon you use.
"It sounds like you know a bunch of *NIX admins who try to run an Exchange server with little or no training, and don't understand at all how Exchange is architected."
My people use native Exchange RPC protocols via Outlook, and then use IMAP or POP3 from PDAs/phones/non-Windows boxes. They even use WEBDAV over HTTP from newer Windows Mobile clients. And yes, everything stays in synchronization, and nothing has ever crashed.
I have no idea what you're talking about with running "only" Exchange. IMAP & POP3 are just protocol stacks that run as part of a functioning Exchange server. From what you say, it sounds like you're trying to sync two mail servers using IMAP... WTF? Neither IMAP or POP3 as protocols were designed for that at all.
Why not just connect directly to the Exchange server via IMAP? Store everything there (clustering if you want/need to deal with those hassles).
It sounds like kown a bunch of *NIX admins who try to run an Exchange server with little or no training, and don't understand at all how Exchange is architected.
The purpose of a corporation is generally to do pursue (usually economic) activities beyond the scale of what individuals or small groups can accomplish. To do so, coporations need to have some legal rights and responsibilites to operate in society as an entity, as such they are a "virutal individual" in the legal sense. Corporations can be for-profit or not-for-profit. Shielding owners from liability is somewhat inconsequential, and in fact only very specific types of corporations shield their owners from legal or financial liability. See references here and here.
Otherwise you'll spend all your time fighting the fact that MS products won't play nice and can't remain stable serving POP, IMAP, and Exchange
I call BS. I have never, ever had an Exchange server fail to interoperate with any non-MS client via SMTP, POP3, or IMAP. It just works, plain and simple, and it is certainly "stable". Most of our Exchange 2003 boxes have 40-50GB databases, yet perform well. The only time Exchagne ever goes down is when we reboot for an OS-level security patch.
We've been using Exchange since v5.5 in a geographically distributed enterprise, and have always allowed non-Windows users to use the POP3/IMAP client of their choice. We've had basically no issues whatsoever with POP3/IMAP interoperability in nearly 8 years.
Shark and Crodcodile civlization hasn't really changed at all in 100 million years. In fact, it would be highly unlikely that if there is life at all in this nearby solar system, it would be intelligent life.
It seems far more likely that we would posess vastly greater intelligence than any species found nearby, based on the number of intelligent species in our own solar system.
The evidence indicates evolution proceeds in fits and starts, with long periods where nothing much happens at all. For 99.9+% of the history of life on Earth, there was no intelligent life.
We just did a similar project, and concluded that iSCSI was the way to go for complete cross-platform accessibility. We evaluated the contenders, and LeftHand came out on top. EqualLogic was a close second. Both vendors allow you to add iSCSI storage devices in smallish increments, which each add their cache, bandwidth, and processing power to the storage pool.
Nothing from EMC, HP, or the other big boys came close in terms of functionality and scalability at this low end of the market. EqualLogic and LeftHand let you start small and grow incrementally without dumping $50+K to get your foot in the door.
If you're looking for the cleanest-looking, most-readable syntax out there... what about Python? Readability and maintainability was a primary design goal.
Dude, the problem with Windows is certainly *not* the kernel, which is pretty solid architecturally and hasn't had many security and amost no stability issues. The problem is with the rest of the "user-space" parts of Windows, like IE, Explorer, etc. Craploads of vulnerabilities and instability there.
The fact that most home users run as root in Widnows doesn't help, making problems in those user-space programs that much more devastating.
Seriously, a well-configured Windows NT-based system (with stable drivers - always a challenge) almost never experiences a kernel-related crash to a blue screen. In 10 years of experience running thousands of NT systems, I think I've seen an actual kernel or HAL crash that didn't involve a buggy video or disk driver like 6 times, back in the NT 4.0 days.
The WIndows Time Service in 2003 SP1 and R2 seems to be a full NTP server. It clock selection of multiple source servers, and keeps time to within about 15 ms of my real ntpd server for weeks, and polls servers at similar exponentially-decaying intervals. All earlier versions seem to be very much SNTP implementations. There's a Microsoft Technet library article describing all the new Windows time service functionality.
That said, the diagnostics still suck; it only reports connectivity issues and insane time sources. The only real way I can figure to make sure it's not misbehaving is to monitor it with another real ntp server.
I think that I am currently the only person in the US who isn't on AIM/MSN/Yahoo/IRC
I don't currently use any IM, either, as I find it terribly annoying to be interrupted constantly. Want to start a "delayed messaging" network with me? We could come up with a simple text-based message format, and then send the messages to each other's computers, where they will sit and wait until we have time to read them. We'll make millions!
Vista-related changes have been talked about at TechEd for years, and much has been documented on MSDN for nearly as long. Of course things have changed along the way, as they do with software in development.
What this is really about is a bunch of software vendors crying that they'll have to spend more money on actualy software development. They want MS to leave things the way they were, despite the problems with the old model. While MS cannot afford to piss off ISVs, making their corporate customers happy by improving the Vista security posture is more important.
Note to open-source advocates: before posing "I can't do X on Windows because it is closed", search MSDN and you'll discover that you're wrong most of the time.
How is VS + ASP.NET + IIS any better then eclipse/netbeans/Idea + Java
The simple answer is: ASP.net requires less code, and less design-up-front to get something working.
In my experience, J2EE is fine for huge project with lots of design resources and a bunch of proficient Java coders, but it is often a sledgehammer in search of something to hit. It's too big and clumsy for many smaller projcts, and offers little if you want to "start small and scale up". Granted, J2EE's requirements for separating logic into MVC makes enforces better application design in many cases, but it is cumbersome if you need to do something outside that design pattern.
Rails is a nifty development platform, but the non-native threading model and interpreted code make for apps that don't scale without a lot of extraneous CPU and memory on the front end.
It's all about the developers. People use IIS because it serves ASP and more importantly ASP.net. Say what you will about Microsoft, but Visual Stuido is a first-class development environment. Building scalable and functional web applications in ASP.net using the graphical tools in VS is easier than anything I've seen in the LAMP world, with the possible exception of Rails.
Plus, Microsoft's near-suicidal devotion to backwards compatibility makes heavily mixed ASP/ASP.net sites like CDW reasonably easy, probably easier than mixing different web frameworks on a LAMP or Java platform.
Why? The Republican-controlled U.S. Congress will not subsidize the build-out of fiber-to-the-neighborhood. And they really shouldn't. Governments in general do exactly two things well: waste other people's hard-earned money, and blow things up. Often they do both at the same time. Governments can't build roads without lots of graft, bureaucracy, and waste. What makes you think internet access will be any different?
If the U.S. government subsidizes a fiber build-out, it will not be cheap at all, and will cost the taxpayer far more than if private enterprise does it. Governments never do anything efficiently. I think you'd find if you fully allocated the costs of the governmental subsidies for fiber access in Korea/Japan/Sweeden/wherever to each subscriber, the high-speed access would cost hundreds of dollars per user per month.
To paraphrase P.J. O'Rourke, "If you think broadband is expensive now, wait until you see what it costs when it's 'free'!"
Slashdot is filled with anecdotes concerning European ISPs that throttle P2P applications. ISPs know that they don't have the bandwidth to supply all of their users with all of the rated bandwidth; it's simply too expensive based on their upstream bandwidth costs. So that's why they have a TOS which allows them to shape traffic with vague terms of what is "excessive usage".
You can get guaranteed bandwidth from some ISPs, but it costs more. Which is as it should be. Bandwidth is purchased by ISPs based on the 95/5 rule, and constant file-trading makes that 95th percentile much higher. We pay for 10 Mbps of guaranteed unrestricted bandwidth at my company and it costs thousands of dollars per month more than the same ISP's best-effort 6 Mbps DSL service.
Slashdot Mirror
You obviously know little about recording. For starters, the analog sections of your PC's audio system are garbage by studio standards. And proper acoustically engineered space doesn't come cheaply either. Many studio-quality microphones (preamp not included) cost several times what the average PC costs.
All commercial software vendors generally prohibit benchmarking to some extent these days. Yes, Oracle and IBM too. The idea is to prevent "bad" benchmarks by improperly configured setups. Most vendors' service organizations will assist journalists and even end-users with performing standard benchmarks (sometimes for a price).
As for SQL benchmarks, they do exist. And Microsoft SQL Server fares very well (espicially on a price/performance basis). See the non-clustered TPC-C benchmarks for four-socket servers for example. I've never seen a TPC result for an open source database listed. Why? Probably because MySQL and even PostGres can't scale like the commercial DBs, and nobody wants to invest time and money it takes to making a sucessful benchmark submission for a product they can't actually sell.
Note that Oracle has largely stopped submitting TPC results, as they've been getting their ass handed to them by MSFT and IBM so regularly.
Many web-based applications already do something better; my employer owns at least one.
The basic technique is to create a derived database access class(es) that include simple caching logic. The whole app then uses that derived class for DB access.
When the cache-aware class(es) are used, the developer indicates whether or not cached data is desiered, and optionally what the specific timeout should be. When a call to this "caching DB class" is made, the SQL that would be sent to the database is instead first hashed, and compared with the cache. If there is a match, and the cache timeout has not been exceeded, the data is returned directly from the cache, otherwise it is passed on to the DB and the result is cached.
Of course, the dev can always specify a query as not cacheable for transactional data, but common things like drop-downs and permissions look ups are quickly cached by this automatic mechanism. When our vendor implemented similar caching in their app, it reduced the average CPU utilization on our DB server by 40%, and made a noticeable improvement in end-user response times. And the amount of code they needed to change was very small; they only had to add 'cache=YES' to the DB access code in a very few critical areas of the application. But the key is that the whole app is still database driven: there are no static files to update when items like menus change.
Two words: content filter.
Seriously, similar problems were predicted recently with the online streaming of "March Madness" (the national college basketball tournament) here in the USA. As a result, I got approval from Human Resources to pre-emptively block CBS Sports during tournament games. Apparently I wasn't the only network manager one who did this... many of my non-IT friends who work elsewhere called me asking "Hey, CBS is blocked, how can I get around this?"
The end result: networks didn't melt, and the Internet lived.
In any case, the streaming sources are limited for such events. The BBC (or other national media outlet) is likely to be overwhelmed before backbones or even end-user broadband connections become saturated.
There are many private University and even non-academic private/corporate libraries to which the same fair use doctrine applies in the USA.
Google cache is really an archival snapshot of the web, much like the archive.org. It is fair use in that it reproduces what is publicly available in any case, but Google adds higher availability and searchability. The orgininal source is duly credited. It's much like a library saving the most recent issues of a newspaper, or the "news clipping" services that are often provided to Congress.
Anyway, I'm sure Google's lawyers understand the legal implications of their caching service far better than you and I, and yet they have not advised Google to take it down.
Traditional "fair use" doctrine, which protects the right of libararies to create things like card catalogs, or allows them to make microfilm copies for indexing and archival purposes. Google serves as a mostly indiscriminate "card catalog" for the Internet. The problem is, as soon as Google starts discriminating about what they index (other than by following robots.txt directions), they open themselves up to having to comply with a million different laws from a million different communites and nations. Google already started down this slippery slope with the Chinese.
MS, and a great number of others, embraced S/MIME over PGP for email encryption many years ago. And you know what? S/MIME just works in Outlook, and certificate management is about as easy as it can get with a PKI.
Now, why did Microsoft (and Netscape, Lotus, Novell, etc.) pick S/MIME over PGP?
Likely because the PGP web-of-trust model is impossible for non-technical users to understand. The WoT is still quite disconnected after all these years of PGP use. While in theory it scales infintely, in practice it doesn't work out so well. S/MIME works just like SSL, meaning the user doesn't have to worry too much about trust, the computer handles the PKI work.
Also, at the time S/MIME was integrated into Outlook, PGP was text-only, while S/MIME offered HTML and attachment support. Very few programs supported PGP/MIME in any reasonable fashion back in the late 90s, and from what I can see the majority of PGP email use still seems to be text-only to this day.
The above statement is totally and completely false. I have been administering Exchange servers since version 5.0 in the mid-90s, and there has only ever been one kind of mailbox. See my previous statements about the pssibilty that the people you reference may not have any understanding the archicture of Exchange Server.
Tens of thousands of geographically dispersed enterprises that run Exchange in mixed environments would disagree with you. I know admins at plenty of Exchange server shops besides mine, and none of them have ever experienced "instability" due to serving mail via POP and IMAP. By the way, if you could describe the nature of the instability, I would perhaps be more inclined to believe your reports. What happens, exactly?
Nice straw man. A Google for "linux pop problems" yields more than 13 million hits. By your logic, we must conclude that Linux can't be made into a functioning POP Server, no matter what POP daemon you use.
Err... that last bit should be:
"It sounds like you know a bunch of *NIX admins who try to run an Exchange server with little or no training, and don't understand at all how Exchange is architected."
My people use native Exchange RPC protocols via Outlook, and then use IMAP or POP3 from PDAs/phones/non-Windows boxes. They even use WEBDAV over HTTP from newer Windows Mobile clients. And yes, everything stays in synchronization, and nothing has ever crashed.
I have no idea what you're talking about with running "only" Exchange. IMAP & POP3 are just protocol stacks that run as part of a functioning Exchange server. From what you say, it sounds like you're trying to sync two mail servers using IMAP... WTF? Neither IMAP or POP3 as protocols were designed for that at all.
Why not just connect directly to the Exchange server via IMAP? Store everything there (clustering if you want/need to deal with those hassles).
It sounds like kown a bunch of *NIX admins who try to run an Exchange server with little or no training, and don't understand at all how Exchange is architected.
The purpose of a corporation is generally to do pursue (usually economic) activities beyond the scale of what individuals or small groups can accomplish. To do so, coporations need to have some legal rights and responsibilites to operate in society as an entity, as such they are a "virutal individual" in the legal sense. Corporations can be for-profit or not-for-profit. Shielding owners from liability is somewhat inconsequential, and in fact only very specific types of corporations shield their owners from legal or financial liability. See references here and here.
I call BS. I have never, ever had an Exchange server fail to interoperate with any non-MS client via SMTP, POP3, or IMAP. It just works, plain and simple, and it is certainly "stable". Most of our Exchange 2003 boxes have 40-50GB databases, yet perform well. The only time Exchagne ever goes down is when we reboot for an OS-level security patch.
We've been using Exchange since v5.5 in a geographically distributed enterprise, and have always allowed non-Windows users to use the POP3/IMAP client of their choice. We've had basically no issues whatsoever with POP3/IMAP interoperability in nearly 8 years.
Shark and Crodcodile civlization hasn't really changed at all in 100 million years. In fact, it would be highly unlikely that if there is life at all in this nearby solar system, it would be intelligent life.
It seems far more likely that we would posess vastly greater intelligence than any species found nearby, based on the number of intelligent species in our own solar system.
The evidence indicates evolution proceeds in fits and starts, with long periods where nothing much happens at all. For 99.9+% of the history of life on Earth, there was no intelligent life.
We just did a similar project, and concluded that iSCSI was the way to go for complete cross-platform accessibility. We evaluated the contenders, and LeftHand came out on top. EqualLogic was a close second. Both vendors allow you to add iSCSI storage devices in smallish increments, which each add their cache, bandwidth, and processing power to the storage pool.
Nothing from EMC, HP, or the other big boys came close in terms of functionality and scalability at this low end of the market. EqualLogic and LeftHand let you start small and grow incrementally without dumping $50+K to get your foot in the door.
If you're looking for the cleanest-looking, most-readable syntax out there... what about Python? Readability and maintainability was a primary design goal.
Dude, the problem with Windows is certainly *not* the kernel, which is pretty solid architecturally and hasn't had many security and amost no stability issues. The problem is with the rest of the "user-space" parts of Windows, like IE, Explorer, etc. Craploads of vulnerabilities and instability there.
The fact that most home users run as root in Widnows doesn't help, making problems in those user-space programs that much more devastating.
Seriously, a well-configured Windows NT-based system (with stable drivers - always a challenge) almost never experiences a kernel-related crash to a blue screen. In 10 years of experience running thousands of NT systems, I think I've seen an actual kernel or HAL crash that didn't involve a buggy video or disk driver like 6 times, back in the NT 4.0 days.
The WIndows Time Service in 2003 SP1 and R2 seems to be a full NTP server. It clock selection of multiple source servers, and keeps time to within about 15 ms of my real ntpd server for weeks, and polls servers at similar exponentially-decaying intervals. All earlier versions seem to be very much SNTP implementations. There's a Microsoft Technet library article describing all the new Windows time service functionality.
That said, the diagnostics still suck; it only reports connectivity issues and insane time sources. The only real way I can figure to make sure it's not misbehaving is to monitor it with another real ntp server.
I don't currently use any IM, either, as I find it terribly annoying to be interrupted constantly. Want to start a "delayed messaging" network with me? We could come up with a simple text-based message format, and then send the messages to each other's computers, where they will sit and wait until we have time to read them. We'll make millions!
I don't know, Singularity seems pretty ground-breaking to me. Far more interesting than anything I've heard about since Plan 9, anyway.
Vista-related changes have been talked about at TechEd for years, and much has been documented on MSDN for nearly as long. Of course things have changed along the way, as they do with software in development.
What this is really about is a bunch of software vendors crying that they'll have to spend more money on actualy software development. They want MS to leave things the way they were, despite the problems with the old model. While MS cannot afford to piss off ISVs, making their corporate customers happy by improving the Vista security posture is more important.
It's certainly possible, and not too difficult, to write your own performance monitors on Windows that plug into the standard perfmon architecture.
Note to open-source advocates: before posing "I can't do X on Windows because it is closed", search MSDN and you'll discover that you're wrong most of the time.
The simple answer is: ASP.net requires less code, and less design-up-front to get something working.
In my experience, J2EE is fine for huge project with lots of design resources and a bunch of proficient Java coders, but it is often a sledgehammer in search of something to hit. It's too big and clumsy for many smaller projcts, and offers little if you want to "start small and scale up". Granted, J2EE's requirements for separating logic into MVC makes enforces better application design in many cases, but it is cumbersome if you need to do something outside that design pattern.
Rails is a nifty development platform, but the non-native threading model and interpreted code make for apps that don't scale without a lot of extraneous CPU and memory on the front end.
It's all about the developers. People use IIS because it serves ASP and more importantly ASP.net. Say what you will about Microsoft, but Visual Stuido is a first-class development environment. Building scalable and functional web applications in ASP.net using the graphical tools in VS is easier than anything I've seen in the LAMP world, with the possible exception of Rails.
Plus, Microsoft's near-suicidal devotion to backwards compatibility makes heavily mixed ASP/ASP.net sites like CDW reasonably easy, probably easier than mixing different web frameworks on a LAMP or Java platform.
Why? The Republican-controlled U.S. Congress will not subsidize the build-out of fiber-to-the-neighborhood. And they really shouldn't. Governments in general do exactly two things well: waste other people's hard-earned money, and blow things up. Often they do both at the same time. Governments can't build roads without lots of graft, bureaucracy, and waste. What makes you think internet access will be any different?
If the U.S. government subsidizes a fiber build-out, it will not be cheap at all, and will cost the taxpayer far more than if private enterprise does it. Governments never do anything efficiently. I think you'd find if you fully allocated the costs of the governmental subsidies for fiber access in Korea/Japan/Sweeden/wherever to each subscriber, the high-speed access would cost hundreds of dollars per user per month.
To paraphrase P.J. O'Rourke, "If you think broadband is expensive now, wait until you see what it costs when it's 'free'!"
Slashdot is filled with anecdotes concerning European ISPs that throttle P2P applications. ISPs know that they don't have the bandwidth to supply all of their users with all of the rated bandwidth; it's simply too expensive based on their upstream bandwidth costs. So that's why they have a TOS which allows them to shape traffic with vague terms of what is "excessive usage".
You can get guaranteed bandwidth from some ISPs, but it costs more. Which is as it should be. Bandwidth is purchased by ISPs based on the 95/5 rule, and constant file-trading makes that 95th percentile much higher. We pay for 10 Mbps of guaranteed unrestricted bandwidth at my company and it costs thousands of dollars per month more than the same ISP's best-effort 6 Mbps DSL service.