Ed
04-08-2009, 02:21 PM
Article From: http://evilgeniusnews.blogspot.com/2009/04/onlive-brilliant-idea-and-concept.html
OnLive, brilliant idea and concept, possible to implement? (http://evilgeniusnews.blogspot.com/2009/04/onlive-brilliant-idea-and-concept.html)
-----------------------------------------------------
OnLive, brilliant idea and concept, possible to implement?
-----------------------------------------------------
I do not particularly like to talk about the current buzz, just nitpick at what discrepancies there are in ideas and theories, like my Apple and Microsoft rage. Yet Onlive caught my attention as it takes current technologies; remote desktop or VNC and then builds on it to solve further problems. OnLive solves piracy, reselling (problem?), compatibility, patching and equality in online PC games, it also allows much further depth in 'console' game patching and upgrades. OnLive seems to be the saviour of PC gaming and homogenises gaming into a single uniform platform where everyone can finally enjoy gaming with anyone else who is playing a particular game.
Unfortunately OnLive hits the Achilles heel of its plan. Networking.
It may well be possible to provide a top range server gaming machine rack for most of its users in a similar way ISP's do to provide DSL. Admittedly the power, bandwidth, hardware and software cost will be astronomical. Yet it is doable on paper and backed by large game companies can slowly grow to accommodate more users due to its monthly and renting payment system.
But OnLive will always, currently, by stymied by networks. The Internet system is fickle, it can experience DNS errors, timeouts, router breakdowns and latency. All except the last will be avoidable with fail safe systems which back each other up. The last will be THE issue that is being talked about. OnLive may have perfect compression technology to offer 720p 60fps to your screen 9more about that later), but when put in a real situation with maybe 1000 users across the US the idea of latency will come into play.
Currently on a PC or console system you run a local client which relays your input to change it; moving shooting etc. It then sends a mirror to the server you connect to to also update your change to all other players. This means you experience minimal input latency and only experience server latency; the time it takes for other player information to return to you and their player to update on your screen. This is why players with high pings seem to click around the map since the server is unable to receive many updates from them and it up to the net code of the game to fill in the gaps.
This means if you get a sever where everyone has a ping of 50-70ms you experience optimal play, since the human mind reacts at about 100ms so you do not really notice the difference. OnLive will experience all of the server lag and input lag since global networking is no where near reliable, effectively doubling latencies to 100-120. This is the equivalent of a European player connecting to a US server and it is very noticeable. Try it by playing a game like counterstrike of battlefield 2 and join a server located on a far continent to you and feel the pain of not being able to hit anyone.
OnLive tech demo at GDC showed a server 50 Miles away running on a dedicated line to the screens at GDC. Note the word dedicated. When you get your Internet access from your ISP they like to provide about 80% of your quoted amount of bandwidth but ramp the latency up to reduce the total requests per minute on their servers. What OnLive probably have not anticipated is the tactics used by ISP to make it seem they are offering you what they said but are actually squeezing the line the contract is written on.
A way to get round this is to build multiple data centers to ensure everyone gets a similar latency. The US is a big place and to get proper coverage is going to push the cost up even higher. Also they will be unable to offer their serve in the EU or Asia until the fully cover that.
They are limited but what reality can offer, sure on paper it works, perfectly at that, but in real life it is completely random how connections work and you cannot rely on the Internet to give any form of stability.
Also another issue is brought up from relying on the Internet. What if your Internet access goes down. Since you are running games through a pitiful box you are now unable to play any games, multi player or single player. Since you now never truly own any games if OnLive goes down you cannot even access them even if you had Internet access. Of course using that argument against Steam is flawed since they offer offline mode which still allows single player access.
To sum up, OnLive could work, they just need the money, of which many big game or hardware companies might not back them up since they are effectively killing the enthusiast PC market and the console market in one fell swoop. They also need the backing of ISPs to stop their shifty ways and provide the service they offered to their customers ALL the time. They are frankly to implement this properly having to fix every problem with systems and networking. If they do it everyone will love them for it. If they don't they just enter the books of failed ideas and we continue our way on happily.
OnLive, brilliant idea and concept, possible to implement? (http://evilgeniusnews.blogspot.com/2009/04/onlive-brilliant-idea-and-concept.html)
-----------------------------------------------------
OnLive, brilliant idea and concept, possible to implement?
-----------------------------------------------------
I do not particularly like to talk about the current buzz, just nitpick at what discrepancies there are in ideas and theories, like my Apple and Microsoft rage. Yet Onlive caught my attention as it takes current technologies; remote desktop or VNC and then builds on it to solve further problems. OnLive solves piracy, reselling (problem?), compatibility, patching and equality in online PC games, it also allows much further depth in 'console' game patching and upgrades. OnLive seems to be the saviour of PC gaming and homogenises gaming into a single uniform platform where everyone can finally enjoy gaming with anyone else who is playing a particular game.
Unfortunately OnLive hits the Achilles heel of its plan. Networking.
It may well be possible to provide a top range server gaming machine rack for most of its users in a similar way ISP's do to provide DSL. Admittedly the power, bandwidth, hardware and software cost will be astronomical. Yet it is doable on paper and backed by large game companies can slowly grow to accommodate more users due to its monthly and renting payment system.
But OnLive will always, currently, by stymied by networks. The Internet system is fickle, it can experience DNS errors, timeouts, router breakdowns and latency. All except the last will be avoidable with fail safe systems which back each other up. The last will be THE issue that is being talked about. OnLive may have perfect compression technology to offer 720p 60fps to your screen 9more about that later), but when put in a real situation with maybe 1000 users across the US the idea of latency will come into play.
Currently on a PC or console system you run a local client which relays your input to change it; moving shooting etc. It then sends a mirror to the server you connect to to also update your change to all other players. This means you experience minimal input latency and only experience server latency; the time it takes for other player information to return to you and their player to update on your screen. This is why players with high pings seem to click around the map since the server is unable to receive many updates from them and it up to the net code of the game to fill in the gaps.
This means if you get a sever where everyone has a ping of 50-70ms you experience optimal play, since the human mind reacts at about 100ms so you do not really notice the difference. OnLive will experience all of the server lag and input lag since global networking is no where near reliable, effectively doubling latencies to 100-120. This is the equivalent of a European player connecting to a US server and it is very noticeable. Try it by playing a game like counterstrike of battlefield 2 and join a server located on a far continent to you and feel the pain of not being able to hit anyone.
OnLive tech demo at GDC showed a server 50 Miles away running on a dedicated line to the screens at GDC. Note the word dedicated. When you get your Internet access from your ISP they like to provide about 80% of your quoted amount of bandwidth but ramp the latency up to reduce the total requests per minute on their servers. What OnLive probably have not anticipated is the tactics used by ISP to make it seem they are offering you what they said but are actually squeezing the line the contract is written on.
A way to get round this is to build multiple data centers to ensure everyone gets a similar latency. The US is a big place and to get proper coverage is going to push the cost up even higher. Also they will be unable to offer their serve in the EU or Asia until the fully cover that.
They are limited but what reality can offer, sure on paper it works, perfectly at that, but in real life it is completely random how connections work and you cannot rely on the Internet to give any form of stability.
Also another issue is brought up from relying on the Internet. What if your Internet access goes down. Since you are running games through a pitiful box you are now unable to play any games, multi player or single player. Since you now never truly own any games if OnLive goes down you cannot even access them even if you had Internet access. Of course using that argument against Steam is flawed since they offer offline mode which still allows single player access.
To sum up, OnLive could work, they just need the money, of which many big game or hardware companies might not back them up since they are effectively killing the enthusiast PC market and the console market in one fell swoop. They also need the backing of ISPs to stop their shifty ways and provide the service they offered to their customers ALL the time. They are frankly to implement this properly having to fix every problem with systems and networking. If they do it everyone will love them for it. If they don't they just enter the books of failed ideas and we continue our way on happily.