Even my single-core ancient Pentium 4 is faster than my 750 kbit/s network connection
A nonsensical assertion. Instructions/second != bytes/second.
For a computation such as "what is the millionth prime", you'd get the answer faster by going to a faster remote service than you would computing it locally, even if you did it over a 1200 baud modem.
Or, something you probably do more often, a question like "find the most relevant web pages matching these words, from your enormous database".
The applications are things like an Ikea website where you can build your living room, place furniture and see a photo realistic rendering of the outcome without waiting a few minutes would be required on a local machine without a render farm.
THIS!
Once you start thinking like that, potential applications start leaping out at you.
There's also the web site where you write type in a script with camera directions, and it renders it into a speech-synth-narrated machinima (I forget the name). This technology could vastly improve both the quality and the responsiveness of such a site.
The first is simply a client->server connection, or perhaps hosting your data online. This, I think, doesn't need a new name. The old names were working fine.
The only people using this meaning of "cloud", are people constructing a strawman argument.
"Since cloud computing is simply [insert old concept], it's a pointless buzzword for an old concept".
If it's not dynamically distributing tasks on a large cluster of servers, it's not a cloud.
How about instead of inventing words we just use the ones we have? Rather than "cloud" computing we could just call it internet-based computing, because that's what it is.
Yeah, we could do away with all kinds of pesky specific descriptions, if we just call everything that touches the internet "internet-based computing". I mean, what idiot coined the tedious and unnecessary buzzword "World Wide Web"?
"Cloud computing" has a meaning. If you can't be bothered to know what that meaning is, that's your problem.
Clue:
If you put up a web server and I browse it, that's internet-based computing, but it's not cloud computing.
If your web server performs some processing for me (to stay on topic, let's say I send it a POVray scene, and which your server renders), that's internet-based computing, but it's not cloud computing.
If your "web server" is actually a massively parallel server farm, spread across multiple datacentres worldwide, which splits the POVray rendering job across multiple nodes such that it completes in a fraction of the time and can tolerate individual nodes rebooting or dying altogether -- THAT would be cloud computing.
Likewise Amazon S3 is cloud computing rather than just remote storage, because the data you store is smeared (mirrored; cached) across multiple nodes so that node failure is routed around, and reads are fast for clients all over the world.
Managing for capacity will certainly be the difficult part of running a cloud server farm.
It's fairly obvious, that if you build your farm to cope with the peaks, there will be spare capacity during the troughs.
But there are strategies to deal with this. For a start you can soften the peaks with pricing strategies. You could even offer discounts for off-peak gaming.
Plus, you could sell your off-peak capacity for other purposes. For example, a Hollywood animation could be rendered using the spare off-peak capacity on a gaming server farm. Substitute your own favourite long-running parallelisable data processing job.
Appropriately enough, given cloud computing's previous buzzword - grid computing - it's not all that different to the way power companies handle fluctuations in demand.
What I don't understand, even if these young'uns have no knowledge of history, why do they think cloud computing is a good idea? Why would they want to offload all the processing onto some distant central computer, when they have a quadruple CPU sitting right here in front of them?
It's that phrase "central computer" that suggests to me you've misunderstood cloud computing. If there's one "central computer" handling my request, I wouldn't consider that a cloud service. A cloud service is by definition distributed. Don't think "big mainframe in a datacentre". Think "huge datacentre full of servers with dynamically managed roles".
Why [...] when they have a quadruple CPU sitting right here in front of them?
Maybe they don't have a quad CPU, and maybe they don't want to buy one.
People misuse the word "CPU" the way they abuse the word "cloud computing"? Really? I've not heard anyone saying they need to buy a new 1920x1080 CPU, or a new 10 gigabyte CPU for their machines.
Never heard someone refer to the entire desktop case and its contents as "the CPU"? "I plugged the monitor into the CPU, but nothing seems to be happening". I got that all the time when I was in IT support.
Which makes me wonder what the point is here really. OnLive is in testing already. TFA doesn't compare to OL so I don't know why Nvidia's offering is so much better.
But I'm curious why you thought a question about publically available datasets had anything to do with cloud computing? It doesn't look as if you were trolling. So what was it you were misunderstanding?
It's got a very well-defined meaning: performing computing and storing data on an internet-connected server from an internet-connected client.
I disagree. If it doesn't involve large server farms, in which the location of your data/process is arbitrary and ideally diffuse, then it's not cloud computing.
Maybe I'm just controller ethnocentristic. And maybe my opponents are too. But I'll take a mouse and keyboard any day over the Xbox controller
It entirely depends what game you're playing.
Quake needs a mouse and keyboard Geometry Wars needs twin analogue sticks. Street Fighter needs an 8 way digital joystick Samba de Amigo needs a pair of triangulated maracas Guitar Hero needs a plastic guitar... and so on.
All of these styles of games can be played with suboptimal controllers (such as Quake on a console controller) - but they each have their ideal controller.
If you can find a copy of the excellent Grid Wars for PC, the recommended control scheme is to plug in an Xbox 360 controller.
I don't know about that. Sure Doom was nice, but it was the original Quake that had everybody I know crowded around a monitor going "oooh!".
I guess it depends on your age and background.
It was the "photorealistic" mountain backdrop in DOOM! that made my jaw drop. I was literally amazed that you could do that on a home system. After all, I'd been brought up on a BBC Micro where Elite was the peak achievement.
Around the same period, I was similarly amazed by Ridge Racer in the arcades. I had rationalised that by putting it down to expensive custom hardware.
Well, if the PC and Amiga versions of Civilization look the same to you, i guess at the time you found that Atari ST graphics looked as good as Amiga's.
It might depend on whether you saw Civ running in CGA or VGA. My recollection is that VGA was by no means ubiquitous at the time. Whereas the Amiga had 8 bit colour as standard.
Agreed, that's why when Half Life came it was so nice. Story in the game flow.
It's nice, but it's still an interruption to gameplay. Even though you can move around, Half Life uses tricks such as locking you in a room until the in-game-cut-scene plays through.
It also allows you to make a mockery of proceedings, by leaping around like an idiot while an NPC does exposition with a straight face. An NPC that you can't shoot in the face...
The point being, that an ADD type like me finds themselves searching for the skip button at times like these. I don't want to be listening to a script - I want to be playing a game.
I dunno. You're absolutely right, and yet... I think DOOM! was probably the first time I perceived a PC as a proper gaming machine.
I mean, Wolfenstein was impressive, and in retrospect (I didn't play it much) a great game -- but it was very much a matter of "well, we've got this PC for business apps, I can make it play this game". At that time, if you had games in mind when you bought a computer, you got an Amiga. Or a console.
Prior to DOOM!, most decent PC games were available for Amiga / Atari ST, with better sound and graphics. Wolfenstein looked like a poor Amiga game.
DOOM! though, came out just as VGA was becoming mainstream, and sound cards were becoming available and affordable. Most PCs didn't have a sound card, and you'd add one as an afterthought, often to improve your DOOM! experience. It looked *amazing* in comparison to an Amiga game, and that was a first.
OTOH the article's author should still consider the 25 years of non-PC videogaming heritage leading up to DOOM!.
You seem to want a computer analogue of a game of dice.
I would prefer something that hides the numbers and gets closer to how things would be in real life.
Not "The enemy attempts an attack. You have 5/20 chance of blocking him... you fail to block... he will cause up to 4 points of damage... it's 3."
But "The enemy attacks you confidently. You can tell that it will take a lot of luck to block him. He hits you and it hurts badly. You can tell he's a much stronger fighter than you."
Now, a good GM might well narrate the dice rolling in such a way. A computer adjudicator needn't show you the dice or the stats.
In both real life and in fantasy combats, there's usually a way of telling when you're severely outmatched.
Such as: "The enemy has muscles on his muscles". "The enemy is waving a huge sharp sword around as if he was born with it". "His first attack is confident and painful". "Your enemy has a fearsome reputation".
*Not* "your enemy's STR and HP are both higher than your own"
A factor of 5 in [CPU] utilization is huge; the question is, is it realistic ? [...] no amount of aggregation of government web site servers will get you much traffic in the middle of the night.
With the right virtualisation tools and parallel algorithms, you're not limited to serving web pages or doing stuff that requires people to be awake. Batch processing needn't be dead.
Those CPUs could be indexing, running academic simulations, processing large datasets (images, videos, SETI, folding@home etc.). In fact the only reason not to have the CPUs at 100% all the time, is energy efficiency.
Having all those processors in a cloud makes them much more accessible for such purposes.
I bet Google's processors aren't idle for much of the time.
Again hypothetically, if some government agency were to decide Amazon was too big and had to be broken up, I would imagine they'd be very keen to do it by splitting into Amazon Retail and Amazon Cloud Services, with Retail buying storage, hosting etc. from Cloud Services.
What you'd probably then see -- since there'd be smart people in Retail -- would be Retail's developers weakening the coupling between their code and Cloud Services' interfaces, so they they could dynamically move between service providers based on price (and other factors).
neither of these companies would even contemplate paying someone else for a shared cloud service to sustain their entire business.
I think, in a hypothetical world where they didn't already have their own huge clouds for historical reasons, they might. Not for their *entire* business, but for the parts that are amenable to it.
Google, less likely. But for Amazon it seems realistic to me.
Hypothetically, imagine there were no online bookseller in 2009, but someone was selling cloud storage along the lines of S3. Someone like Jeff Bezos has the bright idea of selling books online. Part of that is to maintain a huge database of catalogue info, reviews, images, which it's expected a large number of people will be passively browsing for most of the time.
In these circumstances, he'd be a fool to build his own cloud for this. Someone else has already done the hard work.
Now, with today's technology, you probably wouldn't use a 3rd party cloud for the whole of the business. You'd put the catalogue, that takes most of the traffic volume, there, and you'd put the checkout on your own servers, where you had more control over security. You'd have your own systems managing inventory and order fulfillment (which is where Amazon really excel, in my opinion).
Electricity is a commodity and there is essentially no value to be found for your average company in generating electricity. If you treat all of your supporting functions as commodities then it probably is best to outsource them as someone else can do it just as poorly as you for less money.
Exactly, and the core concept of cloud computing is that [insert IT function] becomes a commodity. For example, storage is a commodity. Web hosting is a commodity.
A few years ago, the big companies pushed essentially the same concept, but their buzzword was "grid computing".
What's currently breaking this model is that interfaces aren't consistent. One can't switch from Amazon S3 to some other storage service as easily as one switches electricity suppliers. (Although there are compatibility layers out there).
So either there's nothing of importance in the documents or the workers aren't doing their job anymore.
Or they've reverted to an ad-hoc system of keeping documents on their local filesystems, and emailing them to each other. It creates problems with versions, and "searching" becomes a social networking exercise (or an email to 'all') -- but if workers find it less painful than Sharepoint, that's what they'll do.
In my workplace there's an official Sharepoint site, and dozens of guerilla wiki servers -- Twiki in some cases, abused Fitnesse servers in others.
Slashdot Mirror
Even my single-core ancient Pentium 4 is faster than my 750 kbit/s network connection
A nonsensical assertion. Instructions/second != bytes/second.
For a computation such as "what is the millionth prime", you'd get the answer faster by going to a faster remote service than you would computing it locally, even if you did it over a 1200 baud modem.
Or, something you probably do more often, a question like "find the most relevant web pages matching these words, from your enormous database".
The applications are things like an Ikea website where you can build your living room, place furniture and see a photo realistic rendering of the outcome without waiting a few minutes would be required on a local machine without a render farm.
THIS!
Once you start thinking like that, potential applications start leaping out at you.
There's also the web site where you write type in a script with camera directions, and it renders it into a speech-synth-narrated machinima (I forget the name). This technology could vastly improve both the quality and the responsiveness of such a site.
The first is simply a client->server connection, or perhaps hosting your data online. This, I think, doesn't need a new name. The old names were working fine.
The only people using this meaning of "cloud", are people constructing a strawman argument.
"Since cloud computing is simply [insert old concept], it's a pointless buzzword for an old concept".
If it's not dynamically distributing tasks on a large cluster of servers, it's not a cloud.
How about instead of inventing words we just use the ones we have? Rather than "cloud" computing we could just call it internet-based computing, because that's what it is.
Yeah, we could do away with all kinds of pesky specific descriptions, if we just call everything that touches the internet "internet-based computing". I mean, what idiot coined the tedious and unnecessary buzzword "World Wide Web"?
"Cloud computing" has a meaning. If you can't be bothered to know what that meaning is, that's your problem.
Clue:
If you put up a web server and I browse it, that's internet-based computing, but it's not cloud computing.
If your web server performs some processing for me (to stay on topic, let's say I send it a POVray scene, and which your server renders), that's internet-based computing, but it's not cloud computing.
If your "web server" is actually a massively parallel server farm, spread across multiple datacentres worldwide, which splits the POVray rendering job across multiple nodes such that it completes in a fraction of the time and can tolerate individual nodes rebooting or dying altogether -- THAT would be cloud computing.
Likewise Amazon S3 is cloud computing rather than just remote storage, because the data you store is smeared (mirrored; cached) across multiple nodes so that node failure is routed around, and reads are fast for clients all over the world.
Managing for capacity will certainly be the difficult part of running a cloud server farm.
It's fairly obvious, that if you build your farm to cope with the peaks, there will be spare capacity during the troughs.
But there are strategies to deal with this. For a start you can soften the peaks with pricing strategies. You could even offer discounts for off-peak gaming.
Plus, you could sell your off-peak capacity for other purposes. For example, a Hollywood animation could be rendered using the spare off-peak capacity on a gaming server farm. Substitute your own favourite long-running parallelisable data processing job.
Appropriately enough, given cloud computing's previous buzzword - grid computing - it's not all that different to the way power companies handle fluctuations in demand.
What I don't understand, even if these young'uns have no knowledge of history, why do they think cloud computing is a good idea? Why would they want to offload all the processing onto some distant central computer, when they have a quadruple CPU sitting right here in front of them?
It's that phrase "central computer" that suggests to me you've misunderstood cloud computing. If there's one "central computer" handling my request, I wouldn't consider that a cloud service. A cloud service is by definition distributed. Don't think "big mainframe in a datacentre". Think "huge datacentre full of servers with dynamically managed roles".
Why [...] when they have a quadruple CPU sitting right here in front of them?
Maybe they don't have a quad CPU, and maybe they don't want to buy one.
People misuse the word "CPU" the way they abuse the word "cloud computing"? Really? I've not heard anyone saying they need to buy a new 1920x1080 CPU, or a new 10 gigabyte CPU for their machines.
Never heard someone refer to the entire desktop case and its contents as "the CPU"? "I plugged the monitor into the CPU, but nothing seems to be happening". I got that all the time when I was in IT support.
Which makes me wonder what the point is here really. OnLive is in testing already. TFA doesn't compare to OL so I don't know why Nvidia's offering is so much better.
Nvidia and OnLive are partners: http://www.onlive.com/partners.html
These announcements are probably related to OnLive.
OK, so you've been modded offtopic.
But I'm curious why you thought a question about publically available datasets had anything to do with cloud computing? It doesn't look as if you were trolling. So what was it you were misunderstanding?
It's got a very well-defined meaning: performing computing and storing data on an internet-connected server from an internet-connected client.
I disagree. If it doesn't involve large server farms, in which the location of your data/process is arbitrary and ideally diffuse, then it's not cloud computing.
"Cloud" is a fairly good analogy for that.
Maybe I'm just controller ethnocentristic. And maybe my opponents are too. But I'll take a mouse and keyboard any day over the Xbox controller
It entirely depends what game you're playing.
Quake needs a mouse and keyboard ... and so on.
Geometry Wars needs twin analogue sticks.
Street Fighter needs an 8 way digital joystick
Samba de Amigo needs a pair of triangulated maracas
Guitar Hero needs a plastic guitar
All of these styles of games can be played with suboptimal controllers (such as Quake on a console controller) - but they each have their ideal controller.
If you can find a copy of the excellent Grid Wars for PC, the recommended control scheme is to plug in an Xbox 360 controller.
I don't know about that. Sure Doom was nice, but it was the original Quake that had everybody I know crowded around a monitor going "oooh!".
I guess it depends on your age and background.
It was the "photorealistic" mountain backdrop in DOOM! that made my jaw drop. I was literally amazed that you could do that on a home system. After all, I'd been brought up on a BBC Micro where Elite was the peak achievement.
Around the same period, I was similarly amazed by Ridge Racer in the arcades. I had rationalised that by putting it down to expensive custom hardware.
Well, if the PC and Amiga versions of Civilization look the same to you, i guess at the time you found that Atari ST graphics looked as good as Amiga's.
It might depend on whether you saw Civ running in CGA or VGA. My recollection is that VGA was by no means ubiquitous at the time. Whereas the Amiga had 8 bit colour as standard.
Robotron does have a story - and it was clearly displayed when the game started.
It's a backstory. But it's not a plot that unfolds as as play proceeds. DOOM! has a backstory too - but nobody really cared about it.
movie-length cutscenes
You say that as if it's a good thing...
played by Mark-effin'-Hamill...
... that too.
Agreed, that's why when Half Life came it was so nice. Story in the game flow.
It's nice, but it's still an interruption to gameplay. Even though you can move around, Half Life uses tricks such as locking you in a room until the in-game-cut-scene plays through.
It also allows you to make a mockery of proceedings, by leaping around like an idiot while an NPC does exposition with a straight face. An NPC that you can't shoot in the face...
The point being, that an ADD type like me finds themselves searching for the skip button at times like these. I don't want to be listening to a script - I want to be playing a game.
Hey, what you do mean no story?
No story to speak of. You just broke the rules! Ssh.
I dunno. You're absolutely right, and yet... I think DOOM! was probably the first time I perceived a PC as a proper gaming machine.
I mean, Wolfenstein was impressive, and in retrospect (I didn't play it much) a great game -- but it was very much a matter of "well, we've got this PC for business apps, I can make it play this game". At that time, if you had games in mind when you bought a computer, you got an Amiga. Or a console.
Prior to DOOM!, most decent PC games were available for Amiga / Atari ST, with better sound and graphics. Wolfenstein looked like a poor Amiga game.
DOOM! though, came out just as VGA was becoming mainstream, and sound cards were becoming available and affordable. Most PCs didn't have a sound card, and you'd add one as an afterthought, often to improve your DOOM! experience. It looked *amazing* in comparison to an Amiga game, and that was a first.
OTOH the article's author should still consider the 25 years of non-PC videogaming heritage leading up to DOOM!.
Exactly. Let's face it, DOOM! is basically first person Robotron. Which is Asteroids with walls.
Those games have no story to speak of, and they're fun to play.
The big problem with stories is, you usually have to interrupt the game in order to tell them.
You seem to want a computer analogue of a game of dice.
I would prefer something that hides the numbers and gets closer to how things would be in real life.
Not "The enemy attempts an attack. You have 5/20 chance of blocking him... you fail to block... he will cause up to 4 points of damage... it's 3."
But "The enemy attacks you confidently. You can tell that it will take a lot of luck to block him. He hits you and it hurts badly. You can tell he's a much stronger fighter than you."
Now, a good GM might well narrate the dice rolling in such a way. A computer adjudicator needn't show you the dice or the stats.
In both real life and in fantasy combats, there's usually a way of telling when you're severely outmatched.
Such as: "The enemy has muscles on his muscles". "The enemy is waving a huge sharp sword around as if he was born with it". "His first attack is confident and painful". "Your enemy has a fearsome reputation".
*Not* "your enemy's STR and HP are both higher than your own"
A factor of 5 in [CPU] utilization is huge; the question is, is it realistic ? [...] no amount of aggregation of government web site servers will get you much traffic in the middle of the night.
With the right virtualisation tools and parallel algorithms, you're not limited to serving web pages or doing stuff that requires people to be awake. Batch processing needn't be dead.
Those CPUs could be indexing, running academic simulations, processing large datasets (images, videos, SETI, folding@home etc.). In fact the only reason not to have the CPUs at 100% all the time, is energy efficiency.
Having all those processors in a cloud makes them much more accessible for such purposes.
I bet Google's processors aren't idle for much of the time.
I should add...
Again hypothetically, if some government agency were to decide Amazon was too big and had to be broken up, I would imagine they'd be very keen to do it by splitting into Amazon Retail and Amazon Cloud Services, with Retail buying storage, hosting etc. from Cloud Services.
What you'd probably then see -- since there'd be smart people in Retail -- would be Retail's developers weakening the coupling between their code and Cloud Services' interfaces, so they they could dynamically move between service providers based on price (and other factors).
neither of these companies would even contemplate paying someone else for a shared cloud service to sustain their entire business.
I think, in a hypothetical world where they didn't already have their own huge clouds for historical reasons, they might. Not for their *entire* business, but for the parts that are amenable to it.
Google, less likely. But for Amazon it seems realistic to me.
Hypothetically, imagine there were no online bookseller in 2009, but someone was selling cloud storage along the lines of S3. Someone like Jeff Bezos has the bright idea of selling books online. Part of that is to maintain a huge database of catalogue info, reviews, images, which it's expected a large number of people will be passively browsing for most of the time.
In these circumstances, he'd be a fool to build his own cloud for this. Someone else has already done the hard work.
Now, with today's technology, you probably wouldn't use a 3rd party cloud for the whole of the business. You'd put the catalogue, that takes most of the traffic volume, there, and you'd put the checkout on your own servers, where you had more control over security. You'd have your own systems managing inventory and order fulfillment (which is where Amazon really excel, in my opinion).
Electricity is a commodity and there is essentially no value to be found for your average company in generating electricity. If you treat all of your supporting functions as commodities then it probably is best to outsource them as someone else can do it just as poorly as you for less money.
Exactly, and the core concept of cloud computing is that [insert IT function] becomes a commodity. For example, storage is a commodity. Web hosting is a commodity.
A few years ago, the big companies pushed essentially the same concept, but their buzzword was "grid computing".
What's currently breaking this model is that interfaces aren't consistent. One can't switch from Amazon S3 to some other storage service as easily as one switches electricity suppliers. (Although there are compatibility layers out there).
So either there's nothing of importance in the documents or the workers aren't doing their job anymore.
Or they've reverted to an ad-hoc system of keeping documents on their local filesystems, and emailing them to each other. It creates problems with versions, and "searching" becomes a social networking exercise (or an email to 'all') -- but if workers find it less painful than Sharepoint, that's what they'll do.
In my workplace there's an official Sharepoint site, and dozens of guerilla wiki servers -- Twiki in some cases, abused Fitnesse servers in others.