I can't help but laugh at such statements. The speed of a database is almost completely determined by table structure, indices, caching and the performance of your storage solution.
None of these are in any ways so incredibly complex that Oracle could do it an order of magnitude better than PostgreSQL. So I can only conclude that in these tests one of the systems wasn't correctly set up, missed essential indices or optimized the query in an unexpected fashion.
Any database which has the ability to use indices can easily work with tables with millions of records, as long as the results are a small subset and the query actually uses the index properly (a seek, not a scan of the index). For large result sets, things like clustered indices, caching and the speed of your harddisk subsystem starts to play a large roll. For result sets >10% of the table, most systems will simply perform a table scan (or index scan depending on which columns you actually need).
Of course there are always things that one database can do with good performance that an other can't because of a lacking feature -- in the general case however there's no reason that any sufficiently advanced database system is orders of magnitudes faster than any other.
The biggest crypto traffic is far more likely to be business traffic, like communication between different locations of the same company, access to central servers, VPN traffic, and so on.
The amount of encrypted traffic will only grow; it is practically no extra burden on the systems to encrypt traffic, and for most applications its more lazyness and convience to use unencrypted traffic. I have no doubt that encrypted traffic will soon be more common than unencrypted traffic.
Unlimited bandwidth is most definitely viable. A lot of providers in Holland simply have no limits or donot enforce their "fair use policy". The first few months, you'll be downloading 200 GB every month (I peaked at around 300 GB for a couple of months in a row). I managed to keep that up for about a year orso...
Then I started noticing that I simply could not keep storing 200+ GB of data every month, let alone "consume" it all -- 200 GB is like 400 hours of video, meaning I need to watch video 10 hours or more each day -- I simply donot have that much time. Nor do I have the capacity to store all that much data.
So, now I'm down to like 25-50 GB a month (not including 50-60 GB of upload).
No, just accross the border (Holland) we have no download/upload limits whatsoever (see xs4all for example). Not only that, but its cheaper and faster as well. In Belgium there are only 2 possible choices for a provider, but in Holland there are far more, and there's a lot more competition. Competition is good:)
I'm not going to give away my bandwidth for free for a PAID system... Regular P2P works because giving away bandwidth gets you something of value. Commercial P2P will never work, unless perhaps you get something in return for offering your bandwidth to others.
I wouldn't be so sure, cause once Mom & Pop find out they can't rip their CD's anymore and play them on their computer (or some such) they'll come to me, and I'll explain why this problem suddenly exists. They'll then ask me if I can fix it so they can still play their music, browse the internet and read their mail.
If Linux comes just a tad farther than it currently is on the Desktop (I'm looking at most recent Ubuntu for the current "state" of the linux desktop) I probably will be able to get away with installing it on my parents computer -- I'm sure they'll thank me for all those cool "card" and mayong style games they'll suddenly have at their fingertips...
I suppose it could swing both ways, unless at some point the government interferes, which I'm sure won't be in favor of Open Source (well, not in the US anyway).
Of course not. If 5% of the market is Linux, and that 5% want to continue running linux, they'll want some hardware to run it on. Now, a 5% market share in, say, Motherboard hardware (and a pretty exclusive one too, pretty much a monopoly for Motherboards that STILL work with Linux) is a HUGE deal. If Asus, Abit or whoever doesn't jump on it and produce "compatible" motherboards, then you can bet your ass someone else will.
Or maybe the stupid automatic refreshes on search windows?
Or perhaps the idiotic "Personalized Menu's"?
It sounds to me that it is yet another feature that will get in the way more than that actually helps you -- I don't like it when my machine starts doing all kinds of stuff (with the harddrive) when I'm not using it for 5 minutes.
If you want to start your applications fast, here's a tip: get 2 GB of memory, turn off your swapfile and donot ever close your apps.
Any login system which involves some feedback (like randomized icons you need to select, questions involving your password), passwords using multiple devices (mouse and keyboard), voice based systems, rhythms typed on a single key, and those are just the "quick login" options.
If logging in was allowed to take a few minutes, I could think of a few more -- since I only login every time I need to reboot my system (a few times a year), that's quite acceptable. For example, let the login system ask for 10 random characters in your password ("what's the 3rd character in your password?" or "type 'hjXghkerg', but replace the X with the 5th letter of your password"). Only after the 10 questions are over will the system determine if the login is succesful.
Tell me, would you? Would you spend, say, a million dollars to create something and then give it to the world for free?
Let's assume that we can take car, and make a copy of it (although it will still cost you about $2000 in raw materials). Let's also assume you could make a copy just using the "raw bistream" (or model) without having the car physically there to copy it. Now let's assume EVERY car company is horrified and decides to stop producing cars from that day forward.
I see golden business oppertunities now -- while the old car monopolies are still crying over all the lost revenue, I start a company that creates custom designed cars -- I could create (or buy) a nice model, then hire people to customize these models for customers with special needs. You want a fifth wheel? You want a different dashboard layout? 6 gearbox instead of 5? A 2nd steering wheel? Special needs for specific handicaps?
I could offer support options, perhaps it needs a special type of fuel, or maybe the automated pilot software needs a monthly subscription to stay up to date of the road layouts...
I could even give it away for free to create a lot of free advertising -- and then focus on my core business instead (which may not have anything to do with cars at all). Kinda like Walmart selling the hit CD's at bargain prices to attract people to their stores -- they don't care about the other stuff out there, they just want to lure you in -- very annoying if selling CD's is your core business, but that's how it works.
Also, you'd be amazed what people will do for free, as long as it is fun (for them) and challenging to do. If people could create a new car model, and then create it (physically) you bet your ass TONS of people would do that JUST to drive their self designed dream car -- and some of those will give those designs away... free.
For example, I designed my own filesystem some years ago -- it was very enjoyable, and a lot of fun to get it working fast and correctly, and to add new features to it. See http://sourceforge.net/projects/smartfilesystem (or look on wikipedia) -- it's a project I spend almost 2 years on, then gave away for free.
Surely you can see that from YOUR perspective, however, there is NO difference here. The end result is the same. You have aquired something of value and not paid for it. Moreover, you aquired it at my cost in both cases.
Oh, but there is a difference even from my perspective. First of all, I used my own resources to make a copy. Second, by not physically taking your CD, you are still my friend, and you will not feel obliged to report this to the authorities.
Why? Because you have increased the supply of that item, which conversely decreases the price I can demand for it. As more and more people copy it, it's value becomes less and less. Eventually, in terms of VALUE, my CD becomes WORTHLESS in terms of monetary value, in spite of the fact that I still physically possess it.
I've got legal CD's here, and the last thing I care about is the 'monetary value' of those CD's. In fact, if I told friends of mine they cannot copy my CD's because by doing so it would reduce the value of the things I paid for, they'd laugh.. and rightly so. Not to mention they'd likely will start avoiding me for being a complete nutcase.
In the case of ME, I suppose you could argue that the monetary value of my CD isn't important, it's the personal, or sentimental value I attach to the music that has value, and that measure of value has not changed by other people copying it. True enough.
Yes, that makes a lot more sense.
But this is not true for the guy who MADE the CD. It cost him a lot of money to make that CD. For him, it's the ability to SELL the CD - the MONETARY value that has value. And when you GUT that ability, you have gutted his ability to pay for the process that allowed him to make the CD.
Sorry, I see no reason why an artist that creates a hit-single has the RIGHT to demand payment for that single act millions of times. In fact, that artist would probably never need to create a hit-single again, and could just retire and never do anything artistic ever again -- a great loss for us all.
If instead they were only rewarded once (like a painter for a great piece of work) then there'd be incentive to produce more art. Just because music is so easy to duplicate does not mean the artist should be given a percentage of that for each copy. (Although I suppose, that is already true, since the record companies are usually getting that, not the artist).
However my main reason I will not buy CD's or DVD's but just rip them off whenever I can: I donot like the fact that 95+ percent of the retail price of such an item does not actually go to the artist, but is instead wasted on distribution chains, record executives, marketing, the "music video" (that I will never see, or even want to see), and all that other crap.
The fact that I could get the "goods" easily for free makes a pretty good argument that that other 95% of the price is just one thing: overhead, or more accurately leeches trying to make a dime (and being very good at that) at the expensive of the people producing the actual art -- if you can call the musical crap being produced these days "art".
I've been thinking about that a lot, and I wonder what would happen if I simply give the wrong password (it's the same as the one I use for my ISP, your honor). Then when they only find random garbage, I explain that about 3 months ago, I got wind of myself being under investigation. And thus, 3 months ago (before I even got arrested and before my equipment was seized), I took the precaution of randomizing all of the formerly encrypted partitions.
1) There's no way to prove that the data on the disk isn't random, unless you can find the real password.
2) Would this count as destroying "evidence" if the evidence was destroyed MONTHS before I even was a suspect? Even if I said that I got scared (3 months ago) and decided to go "clean" and destroy it all?
The key points are that, according to me, the drives were randomized WAY before I was a suspect, and that I *did* give the "correct" password, it just doesn't unlock anything anymore...
Funny, I stopped using swap space since I had 1 GB of memory in my computers (2002 I think?). I find it makes them more responsive to the things *I* think are important (instead of the OS swapping out everything, so it can use 999 MB of your memory for cache).
I will consider using swap space again when the following conditions are met:
1) I can limit file caching to a percentage of memory of my choosing.
2) Data that is read at a speed FAR lower than my harddisk top speed (for example, at 100 kB/sec) should only be very shortlived in the cache, if cached at all (this is the main beef really -- I often find that over night I can upload over 1 GB of data, which is all cached -- but there was never any reason to keep it all cached since the harddisk can get to that data much faster than that I can upload it anyway -- the next day however, it can take a few minutes before all the apps I left open are 'snappy' again).
3) I actually run out of memory for my daily use -- with (now) 2 GB of memory, I can tell you that not even extensive Photoshop work will consume all memory (even though Photoshop likes to complain that I have no swap space... as if that would make a difference...)
There's no need to encrypt the entire system, you can just use a boot partition which can decrypt the other drives using a USB key or passphrase (or for the really paranoid some other form of input that cannot easily be logged or taken from you).
My system for example uses several different encrypted partitions with random passwords. There is one master partition (also encrypted) which contains a script to decrypt all the others. This is the only partition I manually mount; it is immediately unmounted after it has decrypted the others. The boot partition is unencrypted, and is fairly small -- I also make sure no comprising information is stored on this partition (like logs involving the encrypted partitions, index files, and so on).
As for plausible deniability that I cannot remember the password, I've come up with a few schemes (although have yet to implement one). The simplest scheme would involve the passphrase getting destroyed if you physically move the computer; something simplistic like a stack of 20 dice in the computer, which would topple if it is moved -- it would be very hard to determine the correct order, especially if they are enclosed in a plastic tube so dust tracks cannot help you either.
It doesn't actually have to contain the password, I just have to make it plausible that it did:)
I can think of several ways around keystroke loggers that would not involve USB keys, or other external storage. You just need to be a bit more paranoid:)
Really, this may go down in America, where they invented the term "Patriotism" (people look at you funny you here if you call yourself a patriot of your country where I live -- kinda like some weird zealot that has too much free time on his/her hands).
ISP's here are FAR more free then what you seem to be used to. They don't block anything, they fight when asked to hand over information, they support Linux(!!), they allow you to run servers, you can turn services on/off at your leisure, and so on.
However, even if ISP's start doing these silly things (which will costs them tons of customers) then there will be other means of communicating digitally. Something similair to the early FIDO/USENET days, or perhaps a network of WIFI'd computers. Or more likely, ISP's that see a new oppertunity and simply provide regular access.
Websites requiring DRM will go the same way as websites using Flash -- if the website doesn't work (or simply sucks, or plays stupid music, requires ridiculous plugins) I don't go there.
I won't be alone, and there will be people catering to the needs of people like me, as I will for them. We might be a minority, but as a Linux user and former Amiga user, I know what that is like -- it's actually not that bad, we never needed Microsoft, or any of the other big guys to get software for or machines, or even custom made hardware.
Nothing short of a government mandate that everybody is required to use Microsoft software as their OS and means of connecting and browsing internet will stop that. Other vendors will literally JUMP at the oppertunity to provide their services to you.
It's the same thing with the music industry. If suddenly all the RIAA company's were shot to Alpha Centauri, music won't die... no, instead there suddenly will be a void BEGGING to be filled, and they'll be standing in line to sell you their music.
You haven't stolen/my/ car, but you have stolen the car from Ford. Why? Because the end result would be the same as if you stole it: You would be driving around in a car that you did not pay for. That car design cost millions to come up with. If they only sold one and the rest were "free copies", they would never make their money back, and consequently, they would never design another car!
Oh they would, or someone else will.
What scares me the most though is that if ever a device is invented that actually COULD copy a car, at will, with the possibility of eliminating hunger, poverty, etc.. that it will be banned because FORD cannot make a profit anymore...
There are some much easier "attacks", one of the primary components of these attacks is not caring that you get an exact digital copy. If I REALLY cared about getting an exact perfect copy, I would store all my data as.WAV files and uncompressed Video... but I don't. Some "noise" introduced by using the analog hole will probably be easy enough to keep small enough compared to the noise and artifacts introduced by the compression to MP3 or MPEG.
So how will the future look in 10 years from now?
There will be lots more bandwidth on the internet for one thing, making it possible for fully encrypted P2P "sub-internets" that do their own routing at speeds likely exceeding current hard disk speeds.
The latest blockbusters will STILL appear on these networks, probably before they're even released in the cinema, like they are now. The only difference MIGHT BE that they are analog "cam" style copies, although still significantly higher in quality than current DVD's. Only one person in the world has to take the effort to create such a copy (and making it harder only makes this process more profitable) for such an unprotected copy to be distributed world wide.
For me it is the biggest hype there is, and having seen (in 2004) a perfectly good web application being completely ruined because the marketing department thinks it would be "bitchin'" to have tabs, sortable grids with draggable columns, checkboxes that change content when they're checked, a window manager and a menu bar (inside the web window) (trust me, it was totally insane).
Not only could we only support Internet Explorer (because there were so many issues that work slightly different in Firefox), but IE would often crash if you used the web-app to intensively (consuming over half a gig of memory wasn't that rare either). Back button was totally ruined, browser history was useless, bookmarking was something we had to MAKE work for every page, same goes for printable views...
Not only that, but server load increased by atleast a factor of two. The application was making so many call backs (typing letters in a field resulted in a call back so it could you show you the closest matches) that we could only support half the number of users on the same hardware. Stuff that used to be "fill in the form" [send request] "fix errors" [send request] became 30+ requests because of dynamic changing layout, content assist, etc..
In the end, it all comes down to what you want. If you want a light weight app, that requires no install and runs on anything, then stick with simple HTML with the occasional java script.
If you want a fancy windows application with trees, editable comboboxes, grids, drag&drop, then write a Windows application (or Java SWT / Swing) and make it communicate with the server.
If you want both, you'll end up with neither; AJAX is not lightweight, nor will it EVER look as good and work as smoothly as a real Windows app.
AJAX is more of kludge than anything; real progress will have to come from an extended HTML spec which will allow for more types of controls and more interactivity.
We can get 6 MBit no cap for 50 euros, and xs4all has full coverage as far as I know, so why he switched to tiscali is beyond me.
I'd never switch from xs4all. They offer unlimited bandwidth (even in their fair use days), allows you to run servers or connect as many computers as you wish (we had a company of 100 people running on a few of their "home" connections), they give you a fixed IP, the line drops literally once a month at the most, bandwidth is always what they advertise (I never had any congestion problems), support Linux, etc..
They're not the cheapest ISP, but I doubt any of the others comes close to their reliability or service.
Slashdot Mirror
None of these are in any ways so incredibly complex that Oracle could do it an order of magnitude better than PostgreSQL. So I can only conclude that in these tests one of the systems wasn't correctly set up, missed essential indices or optimized the query in an unexpected fashion.
Any database which has the ability to use indices can easily work with tables with millions of records, as long as the results are a small subset and the query actually uses the index properly (a seek, not a scan of the index). For large result sets, things like clustered indices, caching and the speed of your harddisk subsystem starts to play a large roll. For result sets >10% of the table, most systems will simply perform a table scan (or index scan depending on which columns you actually need).
Of course there are always things that one database can do with good performance that an other can't because of a lacking feature -- in the general case however there's no reason that any sufficiently advanced database system is orders of magnitudes faster than any other.
...does MS also think it will take them 10 years to catch up?
The amount of encrypted traffic will only grow; it is practically no extra burden on the systems to encrypt traffic, and for most applications its more lazyness and convience to use unencrypted traffic. I have no doubt that encrypted traffic will soon be more common than unencrypted traffic.
Then I started noticing that I simply could not keep storing 200+ GB of data every month, let alone "consume" it all -- 200 GB is like 400 hours of video, meaning I need to watch video 10 hours or more each day -- I simply donot have that much time. Nor do I have the capacity to store all that much data.
So, now I'm down to like 25-50 GB a month (not including 50-60 GB of upload).
No, just accross the border (Holland) we have no download/upload limits whatsoever (see xs4all for example). Not only that, but its cheaper and faster as well. In Belgium there are only 2 possible choices for a provider, but in Holland there are far more, and there's a lot more competition. Competition is good :)
I'm not going to give away my bandwidth for free for a PAID system... Regular P2P works because giving away bandwidth gets you something of value. Commercial P2P will never work, unless perhaps you get something in return for offering your bandwidth to others.
If Linux comes just a tad farther than it currently is on the Desktop (I'm looking at most recent Ubuntu for the current "state" of the linux desktop) I probably will be able to get away with installing it on my parents computer -- I'm sure they'll thank me for all those cool "card" and mayong style games they'll suddenly have at their fingertips...
I suppose it could swing both ways, unless at some point the government interferes, which I'm sure won't be in favor of Open Source (well, not in the US anyway).
Of course not. If 5% of the market is Linux, and that 5% want to continue running linux, they'll want some hardware to run it on. Now, a 5% market share in, say, Motherboard hardware (and a pretty exclusive one too, pretty much a monopoly for Motherboards that STILL work with Linux) is a HUGE deal. If Asus, Abit or whoever doesn't jump on it and produce "compatible" motherboards, then you can bet your ass someone else will.
Or perhaps like the background indexing service?
Or maybe the stupid automatic refreshes on search windows?
Or perhaps the idiotic "Personalized Menu's"?
It sounds to me that it is yet another feature that will get in the way more than that actually helps you -- I don't like it when my machine starts doing all kinds of stuff (with the harddrive) when I'm not using it for 5 minutes.
If you want to start your applications fast, here's a tip: get 2 GB of memory, turn off your swapfile and donot ever close your apps.
If logging in was allowed to take a few minutes, I could think of a few more -- since I only login every time I need to reboot my system (a few times a year), that's quite acceptable. For example, let the login system ask for 10 random characters in your password ("what's the 3rd character in your password?" or "type 'hjXghkerg', but replace the X with the 5th letter of your password"). Only after the 10 questions are over will the system determine if the login is succesful.
I see golden business oppertunities now -- while the old car monopolies are still crying over all the lost revenue, I start a company that creates custom designed cars -- I could create (or buy) a nice model, then hire people to customize these models for customers with special needs. You want a fifth wheel? You want a different dashboard layout? 6 gearbox instead of 5? A 2nd steering wheel? Special needs for specific handicaps?
I could offer support options, perhaps it needs a special type of fuel, or maybe the automated pilot software needs a monthly subscription to stay up to date of the road layouts...
I could even give it away for free to create a lot of free advertising -- and then focus on my core business instead (which may not have anything to do with cars at all). Kinda like Walmart selling the hit CD's at bargain prices to attract people to their stores -- they don't care about the other stuff out there, they just want to lure you in -- very annoying if selling CD's is your core business, but that's how it works.
Also, you'd be amazed what people will do for free, as long as it is fun (for them) and challenging to do. If people could create a new car model, and then create it (physically) you bet your ass TONS of people would do that JUST to drive their self designed dream car -- and some of those will give those designs away... free.
For example, I designed my own filesystem some years ago -- it was very enjoyable, and a lot of fun to get it working fast and correctly, and to add new features to it. See http://sourceforge.net/projects/smartfilesystem (or look on wikipedia) -- it's a project I spend almost 2 years on, then gave away for free.
If instead they were only rewarded once (like a painter for a great piece of work) then there'd be incentive to produce more art. Just because music is so easy to duplicate does not mean the artist should be given a percentage of that for each copy. (Although I suppose, that is already true, since the record companies are usually getting that, not the artist).
However my main reason I will not buy CD's or DVD's but just rip them off whenever I can: I donot like the fact that 95+ percent of the retail price of such an item does not actually go to the artist, but is instead wasted on distribution chains, record executives, marketing, the "music video" (that I will never see, or even want to see), and all that other crap.
The fact that I could get the "goods" easily for free makes a pretty good argument that that other 95% of the price is just one thing: overhead, or more accurately leeches trying to make a dime (and being very good at that) at the expensive of the people producing the actual art -- if you can call the musical crap being produced these days "art".
1) There's no way to prove that the data on the disk isn't random, unless you can find the real password.
2) Would this count as destroying "evidence" if the evidence was destroyed MONTHS before I even was a suspect? Even if I said that I got scared (3 months ago) and decided to go "clean" and destroy it all?
The key points are that, according to me, the drives were randomized WAY before I was a suspect, and that I *did* give the "correct" password, it just doesn't unlock anything anymore...
I will consider using swap space again when the following conditions are met:
1) I can limit file caching to a percentage of memory of my choosing.
2) Data that is read at a speed FAR lower than my harddisk top speed (for example, at 100 kB/sec) should only be very shortlived in the cache, if cached at all (this is the main beef really -- I often find that over night I can upload over 1 GB of data, which is all cached -- but there was never any reason to keep it all cached since the harddisk can get to that data much faster than that I can upload it anyway -- the next day however, it can take a few minutes before all the apps I left open are 'snappy' again).
3) I actually run out of memory for my daily use -- with (now) 2 GB of memory, I can tell you that not even extensive Photoshop work will consume all memory (even though Photoshop likes to complain that I have no swap space... as if that would make a difference...)
My system for example uses several different encrypted partitions with random passwords. There is one master partition (also encrypted) which contains a script to decrypt all the others. This is the only partition I manually mount; it is immediately unmounted after it has decrypted the others. The boot partition is unencrypted, and is fairly small -- I also make sure no comprising information is stored on this partition (like logs involving the encrypted partitions, index files, and so on).
As for plausible deniability that I cannot remember the password, I've come up with a few schemes (although have yet to implement one). The simplest scheme would involve the passphrase getting destroyed if you physically move the computer; something simplistic like a stack of 20 dice in the computer, which would topple if it is moved -- it would be very hard to determine the correct order, especially if they are enclosed in a plastic tube so dust tracks cannot help you either.
It doesn't actually have to contain the password, I just have to make it plausible that it did :)
I can think of several ways around keystroke loggers that would not involve USB keys, or other external storage. You just need to be a bit more paranoid :)
ISP's here are FAR more free then what you seem to be used to. They don't block anything, they fight when asked to hand over information, they support Linux(!!), they allow you to run servers, you can turn services on/off at your leisure, and so on.
However, even if ISP's start doing these silly things (which will costs them tons of customers) then there will be other means of communicating digitally. Something similair to the early FIDO/USENET days, or perhaps a network of WIFI'd computers. Or more likely, ISP's that see a new oppertunity and simply provide regular access.
Websites requiring DRM will go the same way as websites using Flash -- if the website doesn't work (or simply sucks, or plays stupid music, requires ridiculous plugins) I don't go there.
I won't be alone, and there will be people catering to the needs of people like me, as I will for them. We might be a minority, but as a Linux user and former Amiga user, I know what that is like -- it's actually not that bad, we never needed Microsoft, or any of the other big guys to get software for or machines, or even custom made hardware.
Nothing short of a government mandate that everybody is required to use Microsoft software as their OS and means of connecting and browsing internet will stop that. Other vendors will literally JUMP at the oppertunity to provide their services to you.
It's the same thing with the music industry. If suddenly all the RIAA company's were shot to Alpha Centauri, music won't die... no, instead there suddenly will be a void BEGGING to be filled, and they'll be standing in line to sell you their music.
What scares me the most though is that if ever a device is invented that actually COULD copy a car, at will, with the possibility of eliminating hunger, poverty, etc.. that it will be banned because FORD cannot make a profit anymore...
2) I copy your CD. End result: You have your CD, I have a copy of your CD.
Surely you see that there is a huge difference here.
Executions would be good, that would probably wake up a few people.
So how will the future look in 10 years from now?
There will be lots more bandwidth on the internet for one thing, making it possible for fully encrypted P2P "sub-internets" that do their own routing at speeds likely exceeding current hard disk speeds.
The latest blockbusters will STILL appear on these networks, probably before they're even released in the cinema, like they are now. The only difference MIGHT BE that they are analog "cam" style copies, although still significantly higher in quality than current DVD's. Only one person in the world has to take the effort to create such a copy (and making it harder only makes this process more profitable) for such an unprotected copy to be distributed world wide.
Not only could we only support Internet Explorer (because there were so many issues that work slightly different in Firefox), but IE would often crash if you used the web-app to intensively (consuming over half a gig of memory wasn't that rare either). Back button was totally ruined, browser history was useless, bookmarking was something we had to MAKE work for every page, same goes for printable views...
Not only that, but server load increased by atleast a factor of two. The application was making so many call backs (typing letters in a field resulted in a call back so it could you show you the closest matches) that we could only support half the number of users on the same hardware. Stuff that used to be "fill in the form" [send request] "fix errors" [send request] became 30+ requests because of dynamic changing layout, content assist, etc..
In the end, it all comes down to what you want. If you want a light weight app, that requires no install and runs on anything, then stick with simple HTML with the occasional java script.
If you want a fancy windows application with trees, editable comboboxes, grids, drag&drop, then write a Windows application (or Java SWT / Swing) and make it communicate with the server.
If you want both, you'll end up with neither; AJAX is not lightweight, nor will it EVER look as good and work as smoothly as a real Windows app.
AJAX is more of kludge than anything; real progress will have to come from an extended HTML spec which will allow for more types of controls and more interactivity.
I'd never switch from xs4all. They offer unlimited bandwidth (even in their fair use days), allows you to run servers or connect as many computers as you wish (we had a company of 100 people running on a few of their "home" connections), they give you a fixed IP, the line drops literally once a month at the most, bandwidth is always what they advertise (I never had any congestion problems), support Linux, etc..
They're not the cheapest ISP, but I doubt any of the others comes close to their reliability or service.
Now if he asked for a solution that was safe, encrypted and fast enough for streaming media...
You run a 4 drive array, without redundancy? That takes guts I suppose...