Server Costs of Running AR Server

[Scott_171]

I am new to AR so excuse newbie question

Does anyone have a rough guideline about the potential costs of running a cloud server with AR content for a large body of gamers?

 

[Tri23]

That is a really vague question, as most things involved in computers involve technical details.
It would be like, "What does a gaming PC cost?", and the reply would be, "That depends on what kind of gaming you want to do."

For example, "AR Content for a large body of gamers", to me that sounds like you want to know what Nintendo pays to host Pokemon Go at various times in the lifecycle.

A quick google search poses no results, except.. "It's complicated".
You are talking about server processing, response, and ISP time. Is there a server admin forum, because you'd be better off asking somebody who is in the cloud industry on server performance / cost and hosting solutions.

BUT WAIT!
I found an answer!!!

https/www.rackspace.com/en-us/calculator

Insert what you need, and it will spit out an estimated price.

 

[bit_user]

That is a really vague question, as most things involved in computers involve technical details.

I agree with this.

I found an answer!!!

https/www.rackspace.com/en-us/calculator

Insert what you need, and it will spit out an estimated price.

This assumes you already know what you need, which is going to be highly specific to the game. The two basic questions are: how much resources (RAM, CPU cores, storage, and bandwidth) are needed to support a single gamer, and then how does this number scale with the number of gamers? If it's a single-player game, then scaling is very simple and linear. But if it's multiplayer, then there's going to be some non-linear scaling, especially as you grow to multiple servers.

Depending on the type of game, you're also going to need distributed hosting, which adds another whole level of costs and complexities.

 

[Tri23]

Oh, it is completely more complicated than that.

Look at Nvidia's cloud gaming technology. As far as we know, this guy wants to create a AR Fighting game that would be like Power Stone for your desk, where all the computing is cloud based and it is free on the Google AppStore.

a) that would be awesome.
b) that would require a LOT of GPU servers.

That said, the guy's other post was in search of AR developers on here. I think we can safely assume this guy does not understand the undertaking of developing AR and may believe it to be similar to making a phone app.

(that may be unkind, but we are also working on limited information.)

What I see a lot in the tech industry is, people with ideas looking for programmers to make them a product that makes the "idea guy" money. In most cases the programmer goes, "Nah, I have enough work", and if they are a skilled programmer, "Nah, I got more $$$ than I can spend."

 

[bit_user]

Oh, it is completely more complicated than that.

Look at Nvidia's cloud gaming technology. As far as we know, this guy wants to create a AR Fighting game that would be like Power Stone for your desk, where all the computing is cloud based and it is free on the Google AppStore.

a) that would be awesome.
b) that would require a LOT of GPU servers.

I think cloud-based rendering isn't an option, for AR. You could try to do some ATW and the depth-based compositing on the client device, but that still requires excellent latency and I'm not sure what cloud-based gaming can reliably deliver to the majority of clients - especially over mobile networks (this is AR, after all).

So, what I was assuming the cloud would be used for is something along the lines of a traditional MMO-style game world. In fact, I recently read about a company touting such a platform, but I forget the name. It's supposed to scale quite well, and avoid the problem of dividing users into different "shards" or "realms".

One good thing about AR could be that you get relatively even & stable geographical dispersion of players. That is, if it's AR in the Pokemon Go sense. That could significantly simplify the server/cloud side of the implementation.

 

[bit_user]

BTW, good developers learn pretty quickly what sorts of projects & people are worth their time. As for more junior developers... well, at least they can probably learn a lot even from the experience of working on a bad project.