OnLive is dead!?


There problem was they badly missed the mark given the number of active subscribers they had and there estimate on how many people possibly would be active at any time. They therefore bought way, way to many servers to cater for the 1600 concurrent users they had. That's an enormous overhead cost when you think about it.


It's just bad management really. Even having 1.5 million active users with 1600 online concurrently at any time should have been a very successfull business if they managed there HW requirements correctly.


1600 concurrent users is a very hard figure to gauge anything from. For example, if you say poeple only played for 45 minutes on average then 1600 concurrent users could equate to maybe 50,000 unique users a day. Then people don't play everyday, so the unique users per week could add up to quite a bit.
 
Last edited by a moderator:
[Their] problem was they badly missed the mark given the number of active subscribers they had and there estimate on how many people possibly would be active at any time. They therefore bought way, way to many servers to cater for the 1600 concurrent users they had. That's an enormous overhead cost when you think about it.
Not really.

Let's assume for the sake of argument that the servers were 2P socket G34 servers using a 16 core 62xx Opteron part for 32 cores per box. Let's also assume that they were dedicating four cores to a gaming VM and four cores to do the video transcoding compression stream of any given VM. That'd give the ability to simultaneous have 4 players on a server before things started to get dicey with 8 players being a realistic max.


The VM slice would be assumed to have a slice of a shared GPU, and probably have more issues with sound rendering interrupting the normal processing stream then running onboard audio on a normal computer.


With 8,000 servers that would allow a peak of 32,000 users with 64,000 users being the absolute max. That corresponds to 1 out of 47 and 1 out of 23 users being active based on the 1.5 million number, aka ability to handle ~2-4% utilization. There's also the issue you can't design a system around the average number of concurrent users, without accounting for being able to handle the peak.


Note: I expect I'm being generous versus what their actual servers were probably capable of, especially as G34 sockets when they likely bought servers were more along the lines of 8 cores and other complications meaning they were setup to maybe handle around 1% utilization. Their average being 0.1% utilization.


Note 2: I tend to doubt the hardware bill for all those servers came out to more then 40 Million USD.

It's just bad management really. Even having 1.5 million active users with 1600 online concurrently at any time should have been a very successfull business if they managed there HW requirements correctly.
In short... you think they should/could have been successful on a Gym membership style business model? In other words a business model that revolves around making money off people that have intentions towards using your service, but almost never do.

When ability to maybe handle 1% utilization involves excessive sunk costs, and overhead, it's rather clear there's much larger problems with the entire business concept. I'd go so far as to say if there weren't larger mismanagement issues then only someone like Amazon or Google with servers using GPGPUs could be successful at it. And then only because it would be to help maximize paid utilization of their data centers.


Otherwise this business concept requires you to somehow set things up so the customers pay for the hardware to VM the equivalent of a relatively modern gaming system, and hardware of a similar or vastly higher level to create a real time compressed video stream of reasonable quality, instead of just buying a system themselves. This in turn being funded by profits off what have to be competitively priced videogames from services not engaging in such nonsense, and a subscription fee that's low enough to provoke interest. How was that ever going to work as an actual business model, instead of creating leading to massive unpaid and unpayable debts, as occurred?


It doesn't take much analysis to see Cloud gaming wasn't and isn't going to work as a independent and profitable business.

1600 concurrent users is a very hard figure to gauge anything from. For example, if you say poeple only played for 45 minutes on average then 1600 concurrent users could equate to maybe 50,000 unique users a day. Then people don't play everyday, so the unique users per week could add up to quite a bit.
In order to reach the 1.5 million number, you would need to have that happen everyday for an average month, while having no users repeat usage during that month, in a ridiculously coordinated fashion.

If you want look at the larger picture that's fine. OnLive started business in March 2010, which means they've offered their service for slightly less then 2 and half years. Now for the sake of argument let's presume that these 1.5 million were a loyal cadre from the start and the 1,600 is the average for the entire 2.5 years. That'd mean the average user sunk 9.6 hours into the OnLive service over it's entire life. Even assuming a game length of four hours that would mean the average user could only have played a single game per year, and not have touched the service again until the next year.


For an individual game that wouldn't be bad, but this is supposed to basically be an online accessible console.
 
Last edited by a moderator:
^ I think you lost me with all that. They are on record as saying they bought way to many servers for the max concurrent users they had. They are also on record in saying they had 1.5 million active users
 
It doesn't take much analysis to see Cloud gaming wasn't and isn't going to work as a independent and profitable business.

Care to post a link to an article you wrote in March 2010 were you already knew that it wouldn't work?


There is a saying in Germany, and I guess each language has it's own, "Hinterher ist man immer klüger".


Google translates it to: "Hindsight is always wiser".


Back then I found the idea interesting. However, it wasn't available in Europe AFAIR.


And you can be sure they developed some techniques/lessons which will help them (or the buyer) in other projects. So it was not totally non-sense.
 
I never thought it would catch on.the tech wasnt there and it offered nothing over a ps3 or xbox.in fact the games ran slightly worse.not enough processing power to display games on large size flatscreen tellys either.stupid.many many people have 32 inch plus tellys for gaming.why buy a system that is inferior to an 8 year device, namely an xbox360.
 
Back
Top