Forum ‹ GameFS ‹ Expanding functionality

[markloika]

First of all, I think GameFS is awesome and I'm going to use it in Wizard Arena (which unfortunately has suffered a few setbacks due to some unforeseen problems). I'm curious if there are plans to expand upon what GameFS offers in terms of having your whole game stored on the system and just letting the loader spread virally. I'm thinking of what Mochi does with their Live Update service. When a user uploads a new version of the game to their servers, the viral version downloads a small patch instead of downloading the entire new swf. That patch system automatically works using a byte array of the original, and the loader loads the differences between the original byte array and the new byte array, and then patches the original byte array, and then adds the loader object to the stage.

I was thinking that if GameFS could be expanded with that functionality, it would become even more awesome than it is now. It would be really helpful for something such as changing the wording in a menu, or other minor changes. That way the entire swf wouldn't need to be downloaded each time because for games with a fairly large file size that becomes somewhat cost prohibitive.

Cheers,
Mark

As a side note I think it's wonderful that the Player.IO feature set is expanding while the base pricing is staying the same.

[Benjaminsen]

Hey Mark

I must admit that I like the idea of pushing patches to minimize bandwidth use from your GameFS account. It is however not on our roadmap currently, so even though it's a great idea, I do not see it happen anytime soon.

You are however not the only that wish to minimize the traffic used from the core servers and I have heard several creative ways of minimizing traffic. So fare the one I like the best is putting all large assets, such as music and images, into the distributed SWF and load the actual game logic from GameFS.

[danielm]

Frankly, I don't see how the tiny loaders can be done right now. You advertise this use case on GameFS page, but when I try to run the numbers, it just doesn't make any sense to me.

Let's say the game is 5mb and is somewhat (but not excessively) popular, with 200k plays a month. If I'm not mistaken, then 200.000 * 5MB is 1000 GB of traffic. That is already the limit of the "Pro" package, and just a bit more success will already require the enterprise package. At $2500, that is plain and simply too much for a game that is not even a huge success. Not to mention that it is basically impossible to predict the spread of a viral game, and paying for overcharge traffic rather than bigger packages in advance is likely to be even more expensive.

I like the idea to update only the game logic, but that is not a complete solution and not quite what you are advertising. It would for instance be impossible this way to distribute upgrades that include content additions.

I guess as much as I want to like that service, I can't really see any reason to use it for this particular use case over the mochi loader.

[Benjaminsen]

Frankly, I don't see how the tiny loaders can be done right now. You advertise this use case on GameFS page, but when I try to run the numbers, it just doesn't make any sense to me.

Let's say the game is 5mb and is somewhat (but not excessively) popular, with 200k plays a month. If I'm not mistaken, then 200.000 * 5MB is 1000 GB of traffic. That is already the limit of the "Pro" package, and just a bit more success will already require the enterprise package. At $2500, that is plain and simply too much for a game that is not even a huge success. Not to mention that it is basically impossible to predict the spread of a viral game, and paying for overcharge traffic rather than bigger packages in advance is likely to be even more expensive.

I like the idea to update only the game logic, but that is not a complete solution and not quite what you are advertising. It would for instance be impossible this way to distribute upgrades that include content additions.

I guess as much as I want to like that service, I can't really see any reason to use it for this particular use case over the mochi loader.

Great points, let me see if I can answer them all.

Player.IO GameFS is a real CDN solution, this means that your files are replicated to hundred of servers all over the world to ensure your game can be delivered to your clients as fast as possible. For that type of service Player.IO is not the cheapest, but by no means the most expensive either. You can compare prices yourself with Amazon CloudFront which does not have automatic version control nor FTP upload.

Therefore, if your only revenue stream is in-game advertisement and your game is not the type of game will people return to repeatedly, GameFS might not be the right match for you. On the other hand, those type of games typically do not require live updates either.

GameFS does however make a lot of sense for larger social game with higher replay values and revenue sources such as micro transactions where constant updates are important to keep people playing the game. Think Farmville or even Everybody Edits or Shell Shock Live.

GameFS does however provide a lot of advantages, even if you make openly distributed, disposable, flash games. Thus a simple way to lower the cost is by distribution as many of the assets as possible directly to the portals.

[Oliver]

What we advertise about Tiny Loaders from the feature page is:

Distribute updatable .swf files
Tiny loaders are tiny .swf files that simply load the current version of a .swf file. They're tremendously helpful for distributing Flash games, because they allow you to instantly update your main game file across the entire internet by just uploading a new version to GameFS.

You can download a tiny loader for any .swf file that you have uploaded to GameFS.

That is, they' small (Tiny!) files that simply load the latest version of the underlying file... I.e, "Tiny Loaders". They do what we say they do, i don't see how that can be considered false advertisement at all.

Also, your calculation is wrong. If you're on the Plus plan, you get 200GB traffic included and 0.2$ per extra GB, thus the cost is:

(0.20*(1000-200)) + 25 = 185$ per month

GameFS serves a much wider usecase than "just load one main swf". For instance, if your game consist of 500 megabytes of assets, it makes sense to have the "main executable" (swf file or whatever) dynamically load individual pieces when you need them, rather than one big 500mb download...

Mochi has a thing that's targetted 100% towards self-contained flash games, and there is absolutely nothing preventing you for using them for this feature...

You can still use Player.IO for other game features if you want

Best,
Oliver