The new forums will be named Coin Return (based on the most recent vote)! You can check on the status and timeline of the transition to the new forums here.
The Guiding Principles and New Rules document is now in effect.
Cloud Computing: Is it really the future of Gaming?
Aegeri
Tiny wee bacteriumsPlateau of LengRegistered User regular
Tiny wee bacteriumsPlateau of LengRegistered User regular
Recently there has been a considerable amount of discussion over the future of video games, notably because this year Electronic Arts and Microsoft have both begun pushing systems that require online connections. These online connections are being justified by something mentioned much more in relation to games: Cloud Computing. So to begin with, let's define what the hell is actually meant by cloud computing, which unfortunately is going to be difficult because there are actual multiple entirely conflicting ideas of what cloud computing means. Sorry.
In one sense, cloud computing refers to the distribution of a task over a network (an example is the internet) to multiple other machines to solve a problem. A good example of this is actually the folding@home project performed on home computers (and actually, I believe the PS3 as well). Here, researchers encouraged users to download a program that would use their machines to solve how proteins folded. This is a classic example of cloud computing, where one machine distributes tasks among a network of linked machines, in order to harness overall more computational power than is allowed by a single machine. To imagine this, think about trying to paint a house by yourself and how much effort that is. It's much easier (and quicker) to paint the same house and quicker, to use multiple people distributing tasks among them such as "Make lunch", "Paint the front", "Paint the roof", "Paint the walls" and such forth. Most importantly, these need not be real "physical" entities but servers or similar created dynamically on the cloud to handle tasks as they are required. This is the classic "scientific" sense of the use of cloud computing but it's not the only way it's used.
"The Cloud" is also used to refer to a device sending data or storing it remotely elsewhere over the internet, instead of locally. A good example is the way the Xbox 360, most apple devices and Steam store many of your games saved games externally instead of locally on their hard drive. Similarly, games can be theoretically played in the same manner with a server elsewhere holding some (or all of the game) then receiving your inputs and then streaming it back to you as if you were playing it on the actual machine in your house. While I can't think of any outright examples, several games on the upcoming Xbox One claim to use cloud computing in this manner: Calculating things like physical lighting models and then sending them back to your Xbox, freeing the machine up for other tasks.
In terms of gaming this has huge implications and considerable advantages. For one thing, less important and non-immediate tasks can easily be transferred to online servers, which calculate these things and then send them back. Consider a large 4x game like Civilization, where you have 15 AI opponents and each turn those 15 AIs go to individual servers in a cloud. The servers put all of their computational power behind them and end up achieving a signifciantly more intelligent feeling AI at a faster rate than your computer it doing it alone. Similarly, intensive calculations like ambient lighting, water physics and other things can be calculated off the machine, then sent back when needed - again freeing processor time and power for more interesting immediate things. This allows you to make ponentially make better looking and simultaneously deeper games, without needing to have more powerful hardware than you already have. Another advantage is "virtualization", which basically means turning physical hardware into a software equivalent, then making it on the fly as the service requirements change. So as a simple example, let's say you have a server for Battlefield with 9 people on it and another server with 8 people on it. With physical hardware providing each server, they are stuck there and have to migrate to a new server to get the player count up. With cloud based servers, as these servers aren't physical and are virtual, the computers can dynamically change the number of servers or merge them as required. So the 9 and 8 player servers can be merged, keeping the player count up and ensuring the machines aren't running more servers than needed. If more people log on, then they can create more virtual servers and re-distribute the load if and as required.
So that's pretty awesome, right!? Hopefully this below image might give you an idea of the sort of uses overall for "cloud" computing.
I took the image from Wikipedia originally, which has a more overall in depth look at some of the ways it is used here if you need a better description. Returning to the original point of the post, Cloud Computing is something that is becoming more and more common when discussing games. Unfortunately, it's nebulously described and how it is supposed to work rather subjective. So to begin with I really recommend this utterly fantastic article on Cloud Computing from Eurogamer:
In Theory: Can the Xbox One cloud transform next-gen gaming?.
I'll return to the Xbox One and the use of cloud processing later, but in short you will have noticed despite all of the incredible potential that many people in threads this gets bought up in - like the industry thread - are sceptical of the term and some of us (including me) have even gone as far to call it "Outright marketing bullshit". So why is that? What's wrong with this wonderful technology? Well, it's simple: It's limited by your internet. As the Eurogamer article explains - and I really want to emphasize it's a real must read for this topic - cloud computing for gaming has a major caveat to it, which somewhat holds it back and that's the speed of your internet connection to the servers. Take the example of a car crashing, your machine sends this to servers over the internet, the cloud computers process it and then sends it back. The latency between your machine sending that request and getting it back can create some very funky and frustrating results. It's rather like playing on a dedicated server for a multiplayer game, where lag interferes and you thought you "Shot" that guy, only for the latency to catch up and he actually blew your head off a second ago anyway. This is why above, a lot of the processes I suggested were very ameniable to cloud computing for games are all tasks that don't have immediate feedback. Things like lighting physics or similar can be calculated beforehand, then used as required while moment to moment actions like determining if an NPC died from a players bullets/explosion/car crash need to be processed on hand to look correct.
In addition to this issue, "Cloud Computing" has a bit of a PR relations issue after Electronic Arts vastly mislead gamers over Sim City. Sim City, as some of you may know, was originally a largely single player experience but the newest one was based on needing to be "Always Online". In order to avoid the usual backlash that comes with making a game require an internet connection all the time, EA said that the servers were not just DRM they were involved in Cloud Computing. Their response when the game didn't work on release and to calm angry fans, was to basically reiterate that the always online was an essential aspect of the game and it couldn't work without it:
Which seems reasonable enough, until the entire thing fell over when people began actually analysing what the server was doing and leaks from EA revealed some major issues.
And then a modder ripped the games code apart and revealed it worked pretty much as normal, barring some regional functionality gone wrong and that the "cloud" was essentially doing nothing at all. Instead of responding to these allegations, EA just clammed up entirely and quitely shuffled the subject under the rug.
In essence, EA had cleverly tried to sell a really cool concept in the potential of cloud computing to trojan horse DRM into a single player game. Naturally this didn't sit well with very many people and unfortunately, a side effect is now we should all be very cautious about claims that games need to do X for "cloud computing". This is where we return to Microsoft, who are making a huge push right now in selling the Xbox Ones online requirements by mentioning "The Cloud". Microsoft have made some pretty astounding claims on the power of Xbox Ones cloud, such as it having "Ten times the power of an Xbox 360 in the cloud" and how Azure is being used to make games better. Of course there is a distinct lack of specifics on this issue and as the Eurogamer article points out, depending on internet connections/bandwidth speed makes this an inconsistent route to get more power. While I am personally sceptical of some of these claims, the base idea of "Cloud computing" is not bullshit and could have massive advantages for Microsoft or anyone else in future.
This is absolutely awesome sounding technology and this thread is not here to shit on it. The above Sim City example is just to explain where the scepticism comes from - not to be the main focus of this technology. There are so many advantages to us (gamers) if developers can learn to use this stuff well that we could get some really amazing games and functionality out of it. I feel it's going to take some time - especially for internet speeds and bandwidth to catch up to the ambition - but it has all the potential you can imagine. You could buy an Xbox One in November as an example, then in five years time even with aging hardware as internet connections/bandwidth catches up worldwide/in your area it could offload more things to online cloud servers. Meaning your games still keep looking better and better, on par with more powerful hardware, due to sharing much of the computational load elsewhere. It could also work this way for PC games or PS4 games or anything really. Anything that would get you to connect to the internet could give you some really tangible benefits in terms of how it could work out, meaning we could have larger more detailed simulations (which was the hope of many with Sim City), bigger more persistent worlds (as your machine doesn't have to be looking over every piece of it constantly) and more detailed graphics/physics as processes get spread out over multiple machines.
Hopefully this is a reasonable summary of what is meant, how it's been used (or misused really) and what we might be able to look forward to in future.
Edit: Further links and reading,
NVIDIA GRID cloud gaming
Playstation 4 Gaikai Streaming
Microsoft AZURE Cloud system, which it's well worth noting is used for far more than just games
How the Xbox One draws more processing power from cloud computing
In one sense, cloud computing refers to the distribution of a task over a network (an example is the internet) to multiple other machines to solve a problem. A good example of this is actually the folding@home project performed on home computers (and actually, I believe the PS3 as well). Here, researchers encouraged users to download a program that would use their machines to solve how proteins folded. This is a classic example of cloud computing, where one machine distributes tasks among a network of linked machines, in order to harness overall more computational power than is allowed by a single machine. To imagine this, think about trying to paint a house by yourself and how much effort that is. It's much easier (and quicker) to paint the same house and quicker, to use multiple people distributing tasks among them such as "Make lunch", "Paint the front", "Paint the roof", "Paint the walls" and such forth. Most importantly, these need not be real "physical" entities but servers or similar created dynamically on the cloud to handle tasks as they are required. This is the classic "scientific" sense of the use of cloud computing but it's not the only way it's used.
"The Cloud" is also used to refer to a device sending data or storing it remotely elsewhere over the internet, instead of locally. A good example is the way the Xbox 360, most apple devices and Steam store many of your games saved games externally instead of locally on their hard drive. Similarly, games can be theoretically played in the same manner with a server elsewhere holding some (or all of the game) then receiving your inputs and then streaming it back to you as if you were playing it on the actual machine in your house. While I can't think of any outright examples, several games on the upcoming Xbox One claim to use cloud computing in this manner: Calculating things like physical lighting models and then sending them back to your Xbox, freeing the machine up for other tasks.
In terms of gaming this has huge implications and considerable advantages. For one thing, less important and non-immediate tasks can easily be transferred to online servers, which calculate these things and then send them back. Consider a large 4x game like Civilization, where you have 15 AI opponents and each turn those 15 AIs go to individual servers in a cloud. The servers put all of their computational power behind them and end up achieving a signifciantly more intelligent feeling AI at a faster rate than your computer it doing it alone. Similarly, intensive calculations like ambient lighting, water physics and other things can be calculated off the machine, then sent back when needed - again freeing processor time and power for more interesting immediate things. This allows you to make ponentially make better looking and simultaneously deeper games, without needing to have more powerful hardware than you already have. Another advantage is "virtualization", which basically means turning physical hardware into a software equivalent, then making it on the fly as the service requirements change. So as a simple example, let's say you have a server for Battlefield with 9 people on it and another server with 8 people on it. With physical hardware providing each server, they are stuck there and have to migrate to a new server to get the player count up. With cloud based servers, as these servers aren't physical and are virtual, the computers can dynamically change the number of servers or merge them as required. So the 9 and 8 player servers can be merged, keeping the player count up and ensuring the machines aren't running more servers than needed. If more people log on, then they can create more virtual servers and re-distribute the load if and as required.
So that's pretty awesome, right!? Hopefully this below image might give you an idea of the sort of uses overall for "cloud" computing.
I took the image from Wikipedia originally, which has a more overall in depth look at some of the ways it is used here if you need a better description. Returning to the original point of the post, Cloud Computing is something that is becoming more and more common when discussing games. Unfortunately, it's nebulously described and how it is supposed to work rather subjective. So to begin with I really recommend this utterly fantastic article on Cloud Computing from Eurogamer:
Let's start with a quick look at what they mean by 'cloud computing'. Before PR and marketing people brought their fancy buzzwords into the equation, 'cloud computing' was known as 'distributed computing'. This means taking a computational workload and distributing it over a network across available processors.
The idea of cloud computing is to create large server farms full of generic processing power and to turn them from one job to another as needed. Microsoft's existing cloud platform is called Azure, started in 2010, and it's been steadily growing in market share against rival products from Amazon and Google. For developers to be able to use cloud processing, they need both a structure for turning game code into jobs and a means to interface with the servers. This game code model has already been pioneered with this generation's multi-core processors. Developers quickly learned to break their games into jobs on PS3's Cell processor and use a scheduling program to prioritise and distribute jobs across available resources.
In Theory: Can the Xbox One cloud transform next-gen gaming?.
I'll return to the Xbox One and the use of cloud processing later, but in short you will have noticed despite all of the incredible potential that many people in threads this gets bought up in - like the industry thread - are sceptical of the term and some of us (including me) have even gone as far to call it "Outright marketing bullshit". So why is that? What's wrong with this wonderful technology? Well, it's simple: It's limited by your internet. As the Eurogamer article explains - and I really want to emphasize it's a real must read for this topic - cloud computing for gaming has a major caveat to it, which somewhat holds it back and that's the speed of your internet connection to the servers. Take the example of a car crashing, your machine sends this to servers over the internet, the cloud computers process it and then sends it back. The latency between your machine sending that request and getting it back can create some very funky and frustrating results. It's rather like playing on a dedicated server for a multiplayer game, where lag interferes and you thought you "Shot" that guy, only for the latency to catch up and he actually blew your head off a second ago anyway. This is why above, a lot of the processes I suggested were very ameniable to cloud computing for games are all tasks that don't have immediate feedback. Things like lighting physics or similar can be calculated beforehand, then used as required while moment to moment actions like determining if an NPC died from a players bullets/explosion/car crash need to be processed on hand to look correct.
In addition to this issue, "Cloud Computing" has a bit of a PR relations issue after Electronic Arts vastly mislead gamers over Sim City. Sim City, as some of you may know, was originally a largely single player experience but the newest one was based on needing to be "Always Online". In order to avoid the usual backlash that comes with making a game require an internet connection all the time, EA said that the servers were not just DRM they were involved in Cloud Computing. Their response when the game didn't work on release and to calm angry fans, was to basically reiterate that the always online was an essential aspect of the game and it couldn't work without it:
Given similarly disastrous launch week issues that affected online games like Blizzard's Diablo 3 and World of Warcraft, and Call of Duty's Elite service, why weren't EA and Maxis better prepared for this, I asked.
"Each game launch is very different and it's not as simple as saying we didn't have enough capacity," Bradshaw said. "A perfect storm of highly technical issues related to MASSIVE demand cannot be anticipated, only addressed. We are committed to making this right."
What Maxis can't do, despite fan requests, Bradshaw said, is simply strip out the always-online architecture of the game.
"An online interconnected world has been part of our design philosophy since day one," she said. "It's the game that we've been wanting to create since SimCity 4 as we've wanted to explore the dynamics between cities as they exist within regions. Real cities don't exist in bubbles; they specialize and trade resources, workers and more.
"With the way that the game works, we offload a significant amount of the calculations to our servers so that the computations are off the local PCs and are moved into the cloud. It wouldn't be possible to make the game offline without a significant amount of engineering work by our team."
Which seems reasonable enough, until the entire thing fell over when people began actually analysing what the server was doing and leaks from EA revealed some major issues.
“The servers are not handling any of the computation done to simulate the city you are playing. They are still acting as servers, doing some amount of computation to route messages of various types between both players and cities. As well, they’re doing cloud storage of save games, interfacing with Origin, and all of that. But for the game itself? No, they’re not doing anything. I have no idea why they’re claiming otherwise. It’s possible that Bradshaw misunderstood or was misinformed, but otherwise I’m clueless.”
And then a modder ripped the games code apart and revealed it worked pretty much as normal, barring some regional functionality gone wrong and that the "cloud" was essentially doing nothing at all. Instead of responding to these allegations, EA just clammed up entirely and quitely shuffled the subject under the rug.
“I’ve analyzed all of the data calls to and from EA servers – all of the APIs, every request for data, and all of the data that comes back,” explains the modder. And in doing so, he’s found some surprising results. “The SimCity servers are not doing any calculations that could not be done on your PC, even for an entire region single player offline mode, let alone just the city you are in.”
In essence, EA had cleverly tried to sell a really cool concept in the potential of cloud computing to trojan horse DRM into a single player game. Naturally this didn't sit well with very many people and unfortunately, a side effect is now we should all be very cautious about claims that games need to do X for "cloud computing". This is where we return to Microsoft, who are making a huge push right now in selling the Xbox Ones online requirements by mentioning "The Cloud". Microsoft have made some pretty astounding claims on the power of Xbox Ones cloud, such as it having "Ten times the power of an Xbox 360 in the cloud" and how Azure is being used to make games better. Of course there is a distinct lack of specifics on this issue and as the Eurogamer article points out, depending on internet connections/bandwidth speed makes this an inconsistent route to get more power. While I am personally sceptical of some of these claims, the base idea of "Cloud computing" is not bullshit and could have massive advantages for Microsoft or anyone else in future.
This is absolutely awesome sounding technology and this thread is not here to shit on it. The above Sim City example is just to explain where the scepticism comes from - not to be the main focus of this technology. There are so many advantages to us (gamers) if developers can learn to use this stuff well that we could get some really amazing games and functionality out of it. I feel it's going to take some time - especially for internet speeds and bandwidth to catch up to the ambition - but it has all the potential you can imagine. You could buy an Xbox One in November as an example, then in five years time even with aging hardware as internet connections/bandwidth catches up worldwide/in your area it could offload more things to online cloud servers. Meaning your games still keep looking better and better, on par with more powerful hardware, due to sharing much of the computational load elsewhere. It could also work this way for PC games or PS4 games or anything really. Anything that would get you to connect to the internet could give you some really tangible benefits in terms of how it could work out, meaning we could have larger more detailed simulations (which was the hope of many with Sim City), bigger more persistent worlds (as your machine doesn't have to be looking over every piece of it constantly) and more detailed graphics/physics as processes get spread out over multiple machines.
Hopefully this is a reasonable summary of what is meant, how it's been used (or misused really) and what we might be able to look forward to in future.
Edit: Further links and reading,
NVIDIA GRID cloud gaming
Playstation 4 Gaikai Streaming
Microsoft AZURE Cloud system, which it's well worth noting is used for far more than just games
How the Xbox One draws more processing power from cloud computing
Aegeri on
+1
Posts
http://www.neowin.net/news/respawns-first-game-titanfall-announced-for-xbox-one-makes-heavy-use-of-cloud
Titanfall by all accounts appears to be an "always online" game. As such, using IaaS, a dedicated server is spawned to handle the game you and your friends are playing. The various mechs and bad guys you face (sometimes multiple dozens of them) are being run on the dedicated server, which is performing the AI calculations, and handling the physics as well, which are sent to the various Xbox ones connected to said dedicated server.
And when the game is over and people leave the game, said server starts handling cloud PowerPoint for an office team somewhere instead. This is why these servers are never going away.
Let's play Mario Kart or something...
You're mixing a few concepts around:
SETI@Home may enter under some "distributed computing" definition, maybe "crowd computing", though I haven't ever seen that one being used. Here your resources are not "on-demand" (you never order as many people as you need), are not highly elastic (you need more work done? well, the users decide if they loan you their processing power), and there is very limited "resource sharing" (you divide tasks, but that's about it.) These are very defining characteristics of actual Cloud Computing.
Grid-computing refers to joining various resources in order to create one whole, bigger resource. This resource is used as a whole and (as generally understood) not divided itself, but can, however, be shared. Imagine you have this 100-machine cluster and various people give jobs to it, and the machine is programmed to divide its time in some way to work for all these requests.
Grid computing itself is all about resource pooling, but it doesn't include Cloud Computing's main allure: virtualization.
The main draw of CC is that you can "create" resources on-demand and only for the time needed: this illusion of infinite resources couple with this "on-demand" characteristic are some of the main definitions used when dealing with CC. There is resource pooling involved between these resources: your game server may actually involve 1 or 2 physical servers, though you'll never know about it because virtualization only presents you with your one, full server. So now your "grid" divides itself virtually into all these individual machines that serve each client's request, and are destroyed as soon as they're not needed. This allows for a far more efficient use of those resources, even allowing more flexibility with priorities set for certain users (while not outright locking the other users away.)
EDIT: Also, there are not that many applications I can think of that don't involve MMO-like features or virtual dedicated servers. One of the few applications which could be done here may be maritime current models, maybe for the next Silent Hunter or stuff like that, in which the cloud could process a model and feed it to you and your machine would then render currents with that information. Since they don't change drastically, this is the type of application I can see it being used to do. Real-time AI? Real-time physics? Not with current latencies, nope.
http://www.neowin.net/news/inside-windows-azure-server-container
Each primary data center (they've got 4 in USA, 2 in Europe, 2 in Asia) is basically just a heavily secured warehouse with dozens of those things stacked all over the place.
It's kinda awesome how modular the thing is at like every level.
I'm a little dubious on this, and how 'essential' it is. Will the PC version scheduled to be released use 'the cloud' too?
PSN: SirGrinchX
Oculus Rift: Sir_Grinch
Unless you have a hacked version, yes.
Just a few thoughts, this was part of my section trying to explain purpose of Cloud computing to gamers that aren't tech heads.
One reason I'm a bit skeptical of cloud computing with the XBO is that in terms of my analogy you are both buying and renting at the same time, which lowers the efficiency of the whole scheme.
While the technical hurdles are greater, something like Onlive makes more sense to me since it allows people to access cloud based gaming with either something they already own or with a cheap appliance.
Maybe someone that knows this stuff can talk about the economics of enhancing millions of what are already powerful computers. Especially since MS get's payed by the month for XBL and not based on the resources used.
(Please do not gift. My game bank is already full.)
As I understand it you don't have specific "Titanfall" servers. You have a massive cluster of computers and a "Titanfall" profile. When it's required that titanfall profile is loaded on to one of those computers.
So essentially providing they keep the profiles (which is pretty trivial compared to hanging on to actual dedicated servers) the 'cloud' support should be there.
It's really clever. I mean, I don't think for a minute it'll make an absolutely massive difference to the "power" of a console, and it's certainly something Sony could start to do if need be (though MS have the advantage of having a system in place) but it's clever.
PSN: SirGrinchX
Oculus Rift: Sir_Grinch
Avoid the trap Sim City fell into where you lie about what the cloud does for your game, but actually SHOW us how this makes our games better.
Until then it's just vague promises and buzzwords, which are very insubstantial things to put your faith in and, more importantly, bet your money on.
If they release some amazing games with unique features that can't be done on the PS4 then great - I'll buy a One when the games appear, and sit it alongside my PS4 and Wii U.
Until then...
Even then people will remember the PhysX stuff where in order to try and sell their cards they had demos that were purposely gimped when running on the CPU.
Basically they have to put out something where people just sort of know that it couldn't be done without the Cloud.
(Please do not gift. My game bank is already full.)
I can see some nice benefits, but they seem to be mostly of the "under the hood" improvements (more complicated calculations allowed for AI/crowds/cloth/lightning), but at the same time you kind of have to have a working, scaled back, solution in case latency is too high, or if you are in offline mode. It also seems like (at the moment) as something which is slightly tricky to code for? It's just not a matter of "download cloud calculations toolbox" and run it?
I predict two other things to make a much larger difference on how we play games:
1) Lower threshold for smaller studios to enter game marketplaces (Sony and Valve seems to be doing the right thing here). Even if most of the indie stuff is trash, ever so often you will get a Braid, Super Meat Boy and such.
2) Cheap, reliable and easy-to-code-for/port VR. I think Oculus Rift have potential to be great if they hit the right price spot...
How much would that type feature actually add to the gaming experience? It would be cool for the game to play itself, no doubt, but where is the tangible benefit for me, as a player?
It would be useful for lighting and some physics and AI that the player isn't interacting directly with. For example, a zombie game could use it to control the AI for the hordes of zombies not close at hand and wouldn't need to be time sensitive.
For heavy lifting, it runs into the problem of everything requiring a bandwidth of less than ten MB/s even in the best of conditions. This isn't that huge of a problem if everything is handled in the cloud and a video stream is just sent back to the player like with Gaikai or OnLive. This isn't a huge problem for when "minor" stuff is being done, but it becomes a problem when a lot of shit is being done in the cloud. This is why I don't believe the bullshit about the "infinite" power of the cloud or claims that the Xbox One will be four times more powerful with the cloud. The bandwidth simply isn't there.
Here, let Microsoft Studios boss Phil Spencer explain:
http://www.eurogamer.net/articles/2012-06-07-microsoft-internet-bandwidth-issues-make-cloud-gaming-a-challenge
Or Eurogamer:
http://www.eurogamer.net/articles/digitalfoundry-in-theory-can-xbox-one-cloud-transform-gaming
Well in Watchdogs specifically you could have Extras that behave in a rational fashion with changing motivations that you run into later, that'd be pretty neat. A GTA you could have stuff like traffic accidents elsewhere causing gridlock, finding the aftermath of random violence just because, probably a bunch of other things I'm probably not able to think of off the cuff. It would just make a genre that thrives on presenting the illusion of a living breathing world feel even more alive.
All of these things could be easily generated when you visit an area.
Stuff like, there's a 25% chance every half hour that the game will pick a certain intersection for an accident to have occurred with backed up traffic. If you get near it, you'll see the effects.
The accident doesn't have to have happened because cars are driving around with simulated AI and one of the aggressive AIs ran a red light while the distracted AI was doing her makeup and didn't notice so they crashed. The player just needs to see an accident every once in a while, and you can just insert an already-occurred accident in the world nearby but out of sight at regular intervals. Far more efficient, easy, and results in the exact same experience for the player.
We discussed this particular possibility for cloud computing at length in the industry thread...there is no real benefit to simulating the entire world.
In fact in a game like Skyrim where NPCs can randomly fall through the world and be lost forver, you wouldn't want this to be able to happen constantly as NPCs wander around miles away. It just exacerbates these problems.
Instead of your game crashing because something weird happens nearby, now it can crash because an NPC tried to do something impossible 50 miles away.
EDIT: Note that of course I am not saying cloud computing is useless for gaming, but I don't think we should focus on uses like this as good examples of what it could do for gaming.
I don't even know what this is supposed to mean.
The server farms are always going to be useful. They can run instances of anything, and they can split off authentication/MP servers indefinitely. There is no reason to ever shut them down.
Microsoft shutting down the servers is not a rational concern under this architecture.
No, they wouldn't shut down the servers. They'd just use them for something else. That's the real advantage of Azure.
Seriously, server farms consume HUGE amounts of power, and cost an enormous amount of money to run, even when all the hardware is in place. Microsoft doesn't run them for funsies. If it's not cost-effective to keep the servers running One stuff, they'll definitely switch them over to running Office 2019 or something.
I wouldn't be surprised if the PC version required dedicated servers. Which is essentially how they are using the cloud for this application. the dedicated server is running the AI, physics and multiplayer stuff for all players in the game.
The XBO has the advantage of having as many dedicated servers as are needed for their playerbase, wheras the PC version will probably require you to set a dedicated server up for your players, or pay a monthly fee to rent a box somewhere.
The 360 version isn't being developed by them, and simply won't have the scope of the next gen version. I am thinking "dead rising" for the Wii.
Let's play Mario Kart or something...
Server virtualization does not work this way
You don't "switch something over"
It's all created on demand and terminated when no longer needed
Or, to make it easier for this guy.
The cloud servers for the XBO are files. Relatively small files at that. These files are sever / computer profiles.
When nobody is using them, they cost microsoft nothing at all except for storage space.
People pay 50-60 dollars a year to have access to online multiplayer. When they want to play a game, the game sends a request to Azure, which uses the file(s) to spawn a dedicated server or a physics processing platform or whatever is requested that is regionally appropriate for the end user. When the game is doing being played, the server deallocates its resources and ceases to be.
This is why they are never going to switch it over to become an office server instead of a Live server. Because it already is an office server, and the same cluster of machines will be able to run whatever task is requested of it at the time of request.
And this isn't magical pixie dust unicorn horseshit. This is how this technology already works and has worked for many years now. On this very platform (Azure). What Microsoft is going that is progressive and neat is applying this proven technology to their online gaming platform, which will drastically reduce costs for offering things like dedicated servers indefinitely, and will radically increase performance for all games that demand online multiplayer, at the very least.
As a note, I am not entirely sold on physics and AI calculations on truly singleplayer games yet. I feel like we are a few years away from needing to offload stuff like that from this hardware; its pretty fucking powerful already. But from a logistical perspective, having a dedicated server handle all of those calculations for a multiplayer game (like titanfall) and distribute it to the connected xboxes means that nobody gains host advantage, and the game will not be gimped by the connection of the host. AND you can still do some great physics and AI.
It's win/win/win.
Let's play Mario Kart or something...
And then you use it for something else. So, by terminating something and then replacing it with something else, you've switched the physical servers over to something else. These virtual servers have to run on something, after all.
And thank you for agreeing with me -- Microsoft will likely terminate the One servers and make room for something else when the One stuff is no longer cost-effective.
@syndalis : Thank you. Honestly. But I seriously doubt Microsoft will still dedicate hardware space to the One until the end of time, especially when those resources could be used for something else. Especially considering server capabilites get shut down all the time even with this technology.
And just to help you along here because I know english is a second language for you...
These are not "ghosts" like you are used to in other racers. They pick up your tendencies, not only on individual tracks but across all races. Do you tend to hug the curves on Nürburgring and accelerate on the downhills, but act rather differently on other courses? Takes it into account. Do you get aggressive when someone tries to pass you and attempt to cut them off, push them into a braking situation? Taken into account. Do you do that moreso in certain track conditions that others? Also taken into account.
What gets created is essentially an adaptable "AI" based upon your driving performances, that improves the more you play the game.
And if your AI wins races, you get paid for that, even if you aren't at the console.
It's a really neat idea. And one that while capable of being done on your console, can be sent off to the cloud for processing without causing you to stare at a 2-3 minute load screen after every race as it calculates.
Let's play Mario Kart or something...
"I think I can comment on this because I used to live above the Baby Doll Lounge, a topless bar that was once frequented by bikers in lower Manhattan."
I honestly can't tell if you're being deliberately obtuse or you really don't understand.
There is no "terminating the servers." The servers are constantly being generated and terminated. When someone runs Office, they're not connecting to an extent Office server - a new one is spun off in software, exists until that user closes the program, and then terminates. There's no cost difference between running Office and One servers and therefore no reason to ever prevent One servers from instantiating, particularly considering that the auth servers are going to be about the most light-weight thing you can possibly run.
Me ? I am an educated Business System Analyst.
Let's say there are 1000 games on the Xbox One. Each of the 400 mil XBOs (I'm going off of Microsoft's numbers here) has an allocated processor/memory/bandwidth space in the Azure cloud equal to.. 4 of them, was it? So 1.6 billion XBOs worth of azure space - easily doable once Microsoft reaches those numbers, mind you. But any time a player launches a game and connects up to the cloud, the Azure needs to take that allocated space, load the cloud computing parts into it, fire up the VM, and present the player with their stuff. This is, of course, assuming that these aren't always persistent VMs, which would never be spun down.
Again... somewhere in that chain is some sort of cost, be it the offline hard drive storage of the reference templates and configs, the persistent cloud, the bandwidth, whatever. Presumably Xbox Live Gold will cover those costs. At least, that's what Microsoft is banking on - the XBO player is just going to become another tenant in the Azure cloud.
... which, if I may be pessimistic, has really sucked for me lately. The last 2 weeks straight there have been problems with Office 365, be it the Exchange servers or the Lync servers deciding to go rogue. When we started moving our customers to it, it had an uptime of 99.999%. I swear it's dropped to 99.99 or 99.9 by now... and it is starting to piss off some of our customers who are trying to figure out why it isn't more reliable to just host locally. (I'm trying to learn more about hybrid deployments, but good god is federation expensive from a licensing standpoint, let alone physical hardware.)
I think the real problem is that you're talking about technology and I'm talking about money. Spawning new Xbone servers wouldn't be worth it if any other service could bring in more money for the same resources. You can't just spawn a virtual server in thin air, they're competing for space.
Besides, as Syndalis said we live in a world of server virtualization and yet services still get terminated all the time due to cost issues. Why would the Xbone be any different?
Ah so you guys use the same phrases as SaaS sales reps. Cool.
"I think I can comment on this because I used to live above the Baby Doll Lounge, a topless bar that was once frequented by bikers in lower Manhattan."
"I think I can comment on this because I used to live above the Baby Doll Lounge, a topless bar that was once frequented by bikers in lower Manhattan."
Then they bitched when we had to manually add printers to every PC.
Then they bitched when their DSL wasn't fast enough.
Then they bitched when their fiber wasn't fast enough.
Then they bitched when it took 14 steps to upload/download any files.
Cloud processing and computer is awesome. I really like what I've seen of Office 13 and how seamlessly it does things, and I hope the XBO is the same way. But there's just too many negatives for me to get behind it to this extent. Maybe if the XBO was $100 w/ a 15-30 Live subscription. Maybe.
That's actually one of the parts that has me the most excited, though dedicating upwards of 2GB to the hypervisor (if rumors are true) seems incredibly overkill-tastic. But yeah. I love VMs, and wish VMware training was a hell of a lot cheaper than it was.
Salvation needs to be specific about his expectations.
From what I am reading with his continued insistence as to how the cloud works, he believes that in 60 years when the antique store owner brings out his vintage Xbox One to show a curious customer, he'll blow the dust off, plug in the system, connect to his slow-ass 1 terabit internet, and Azure will generate an instance to serve him for his little 20 minute demo of how gaming technology used to work.
Except I don't work on commission and I actually do analyst work and architectural design of new system. My company has been leveraging VDI and VM solutions for years now. I do think I have a fairly accurate understanding of its capabilities and potential application.
... I think you just unintentionally answered the question of how a museum would handle this system. No, seriously - museums are finding it very hard to curate modern video games, as they can't replicate the multiplayer and online components. It wouldn't be that hard to give them a dedicated cloud space (or better yet a local server farm) and hand out the experience as best they could, even allowing networking with other museums to help replicate things.