How can Photo Cloud be cost effective for an iOS game?

I'm looking at this here:

http://cloud.exitgames.com/Pricing#

So lets say my mulitplayer app sells 1000 copies at $2.99 I then take home maybe $2000 after taxes. Then if I am an indie I have to give $759 over to photon cloud in order to cover all my 1000 users on multiplayer, and that has to be paid each year. After just two years mostly any money my game made would have already gone to photon.

Or am I missing something here?

Or is photon cloud really more intended for MMO subscription games, or freemium games that have continual income, not just one time purchase games?

 

Mmos earn money by subscriber
there is no mmo sold as a one time purchase

In contrary, most of one time purchase game come with server freely hostable by player or team of players, they aren't hosted by the game publisher

modern fps use p2p techniques to reduce servers costs

 

Learn what CCU means.

 

Photon Cloud is not for MMO at all, it has no persistence nor server scriptability, the two core aspects required for an MMO

As NPSF3000 in a not that friendly way tried to tell you, you need to understand that CCU = Concurrently Connected User means the number of users online at the same time, which is normally at best in the range of 10% of your installbase, often even below that beside the peak point.

So if you sell 5000 copies, you are potentially not even able to make full use of a 500 CCU subscription while you easily make enough money to pay an even higher one.
Also nothing forces you to have an 'all out CCU subscription' all the time, that would kinda defeat the point of a cloud and its scalability if you couldn't adopt it to your needs. Best talk to ExitGames directly on that, explaining what you are after etc

But let me assure you that it is extremely cost efficient. Hosting an own dedicated machine to offer multiplayer services costs you $80+ a month just for the machine rent in an acceptable datacenter and that without all the time required to maintain it, set it up and keep it running while being up to date which easily will cost you many times those $80 a month unless you rate your work time at unrealistic hourly rates

 

I'm with dreamora, talk to ExitGames.

I'm working with PhotonCloud everyday for demos and general testing, it's a good platform with solid connection. Although i would never ship a game using PhotonCloud the way its presented in the web site, paying for fixed CCU nowadays is hard for indie developers like us, i would go for Amazon EC2, paying for what i use.

If you are successful or not in contacting them for a more scalar price please tell us.

 

While I do say that photon are very approachable - so do talk to them - WTF?

If you think that EC2 means 'paying for what you use' you are very sadly mistaken - EC2 means 'paying for what you might use'. Unless your particular business case requires/allows you to take advantage of this model, it can easily be far more expensive than the alternatives.

With photon cloud $9/month will allow you to build and release a game that can handle ~1000 daily players. If you get more than that you can upgrade to the $39 plan for ~5000 daily players and so on. Easy peasy.

EC2 on the other hand has a very different structure. For a start, as soon as you exceed 100 CCU you need to pay $800 for a photon license. Then of course you have to pay for a the EC2 instance - the small windows instance will cost you $82/month for the 'vps' alone. Factor in $120/TB bandwidth, server admin, monitoring, redundancy etc. and it quickly becomes a comparatively expensive proposition.

The best part? Unless photon have stuffed up big time... even after all these costs photon cloud will still outperform your EC2 set-up.

That's not to say the amazon platform isn't without uses - I use amazon for both testing and production - it's not anywhere as easy as many believe.

 

Interesting discussion - just summarizing and adding some insights we got

CCU == Concurrent Users == Payers online at the same moment in time

Based on our experience you can use the following rule of thumb:

MAU = 1000 (Monthly Active Users, in your case 1,000 downloads, assuming all go online)
DAU = 10% * MAU = 100 (Daily Active Users)
CCU = 10% * DAU = 10 (Concurrent Users)

So you can easily start with the 100CCU subscription for $9 a month. With that subscription you can handle 10,000 Downloads and will make $20,000 (using your numbers).

So we ask for 9$/$2000 = 0.5% of your net revenue - we think this is very fair.

Photon Cloud is focused on "room" based type of games: So up to lets say 15 players per room.
(It is not suited for MMOs with 1000 people per region/world - where you need interest management.)

Here are some references using the Photon Cloud:
- racer: http://goo.gl/1XArY (Moboygames, Road Warrior)
- fps: http://goo.gl/KVfRh (Studio on Mars, Critical Strike Portable)
- tower defense: http://goo.gl/AazkS (3sprockets, cubemen)
- coop: http://goo.gl/C7ys6 (Ninja Kiwi, SAS: Zombie Assault 3)
- social realtime: http://goo.gl/Gb67X (xlabz, Sketch with Friends)
- casual: http://goo.gl/SUT8E (Windows 8 Launch title, Windows Phone)

NPSF300 is right: You have the following monthly cost components
a) Instance (you can't turn it off) / this should be around $100
b) Traffic / in our Cloud 1,000 CCUs cause 6.5 TB (!) of traffic
c) License / Photon 100CCU is free, Indie is $39
d) Operations (what is your time worth?)

So for 100CCU you have:
a) $100
b) $78
c) free
d) 50$? 100$?

 

Why limit yourself to Photon?

Host your own server, and get a uLink Indie License for 550 Euro a title.
http://muchdifferent.com/?page=game-buy

That's a once off payment, then all you need is your hosting costs, the rest is pure profit.

 

That's probably the most expensive solution outlined so far

 

Well, it depends. If you're going to have a high CCU count, over a long period of time, then uLink would be more affordable.
What is most affordable depends on your game, and how long you're going to provide multiplayer for it.

 

1) There's nothing really to suggest a High CCU - if anything the OP suggests the opposite.

2) Given that ulink runs inside of U3D, I'd expect it to be less efficient than photon in terms of raw CCU.

TBH ulink and photon have drastically different approaches to solve the same problem, so trying to argue hosting cost [other than the extreme low end, where photon cloud is simple cheap] is pointless - there are bigger concerns.

 

If the OP has only 100 CCU, then great, Photon is free.

However if he has 101 CCU or more, then he has to pay $799 per year. He estimates 1000 sales, dreamora based 10% as a fair for CCU's, but 10% IMO may be a bit low, you don't want to stop players connecting to your game, its bad for your reputation and bad for business, especially if multiplayer is a feature people paid for.

This is where I think uLink is the more cost effective option. You pay your 550 Euro once off for the title, and you're not limited at all. Your game hosting is only limited to your hardware, and this is for the life of the title, depending on which may last a year, less, or much longer if successful.

So aside from only wanting no more than 100CCU's, I don't see how Photon is the better option.

What exactly do you mean by this?

 

Err... it's a once for 'unlimited CCUs Servers' - much like ulink.

Except that 1000 sales probably translates to ~10 CCU [see bertel's post] - so 100 CCU does have a fair bit of legroom.

Because, as you'd notice in this thread, photon will provide a hosted service for far less than your VPS costs - let alone management, licenses etc.

uLink - old version of mono running inside of instance/s of a game engine.
photon - high performance C++ core with .NET 4.0 scripting.

Now call me bonkers... but I think photon*should* be able to deal with substantially higher amounts of traffic per server. I could be wrong though.

I still think trying to compare the two on CCU grounds is fairly bonkers considering the vastly more important differences between the two.

 

That's what I thought, I'm just wondering what is the OP going on then about having to pay the fee each year?

Aaah, you're talking about the server implementation, excuse me I've just woken up and still half asleep lol... that doesn't automatically mean it will be more efficient than the other I think it comes more down to the implementation. But for sure you can't beat perfect C++ with the latest tech.

 

I suggest you spend 10 mins reading up on photon cloud. It has a lot of draw backs, but it also has a couple real nice features.

Nothings guaranteed... but ulink would be missing one hell of an advertising opportunity if their capacity was remotely as good as photons.

 

I originally had Photon implemented in my game. then moved over to uLink. I was basing the whole yearly fee based on the OP's post, I thought that was something new in their licensing model. Perhaps the OP will feel a lot better when they realise this is not the case

 

Add a simple queue system for people logging in.If the queue times are longer than say, 15 seconds, it's probably time to add a bit more space. But it boggles my mind you would go this route for your first game.

 

Suicide, if you've ever run your own server... for a start the latency is horrible a couple thousand miles away, and secondly, it will go down at some point. Hard drives in dedicated servers get hammered much harder than a normal desktop hard drive.

 

I'd rather have control over my own server infrastructure than let somebody take care of that for me. My MTR game will be launching with a server in Central US, one in the Netherlands, and one in South Africa. All the databases and logging etc I'd like to manage myself, and write my own login system. Having your own servers = more finely grained control.

But for the OP for sure, perhaps having Photon Cloud managing everything for him is the way to go, but any users not living close to Photon's cloud infrastructure will have high latency anyway and extending that infrastructure won't be possible, unless Photon extends it, or he hosts his own infrastructure entirely.

Thats why you get a decent host that offers decent servers that don't go down and you get RAID.

I would hardly call good planning and getting your own servers 'suicide'.

 

I just gotta bump this thread to say that, after reading all of this, I think photon would be a lot more appealing if it had a 'bleed' percentage (term I just made up), or rather if CCU's were based on average amount of users, rather than a max count. What deters me from photon is the possibility that at some point my app will have a radical, unexpected, user spike for a short period of time.

Cause the thing that gets me about photon is, ok, so I sell 1,000 copies and this gets me on average 10 CCU. But what happens on some day there is like a sale on the app store, or some holiday, and then by chance I completely blow past that 100 person limit for 1 day? Lets say up to 500 people. That would be a nightmare for reputation, the majority of my users that day will remember the multiplayer as a piece of S***. Then I am supposed to upgrade my account for an entire month just to catch potentially a few days that have 'spikes' in users?

See I don't want to be paying $169 a month, just to be able to cover maybe 4 days out of a month when I have 600 CCU, whereas the rest of the month I barely push 50 CCU, if at all.

I would be much more inclined to use photon if there was somehow insurance set in place in some way to be sure that if my game spikes, it doesn't ruin my games reputation because I didn't get an account with enough CCU. This is why I am thinking, what if the CCU count was based on an average instead of a monthly maximum?

So lets say 90% of the time your game has 10 CCU, another 7% your game has 80 CCU, and then another 3% it has 500 CCU. In that instance you really shouldn't be paying for the 1000 CCU user account, you should still be paying for the 100 CCU account.

I'm thinking if it went like this:

You pay for 100 CCU user account. Photon hosts your game for that entire month, keeping track of average CCU, hosting all players, no matter what.
The following month if you manage to be above an AVERAGE of 100 CCU lets say 20% of the time for the entire month, you would be forced to upgrade to a higher level account for the following month.
If through that month your user base dropped down to below an average of 100 CCU, then you could downgrade your account again.

This keeps it constantly dynamic, and makes it so you don't have to worry about losing users because of a random user spike and not having an account with enough CCU.

I have no idea if that would work, but thats the only thing I can think up to solve the issue of random user spikes.

 

Except of course, that their system, will scale to handle the users without any sort of throttling.

Don't make up fictional scenario's when you could just as easily ASK THEM.

 

But don't you then have to pay more?

 

Yes... and the question is?

 

I think the big question is does the Photon service automatically upgrade one's account if the CCU limit is exceeded? If not, then those 'spike' users will indeed experience failed connections and sufficient time would not be available to upgrade one's account manually. That could be mitigated somewhat if notification of attempts to exceed the CCU limit are immediate and it is very fast to perform an upgrade (at any hour), but it would still result in some number of new users failing to connect. With the proliferation of apps for various systems, the common user is much more likely to delete an app with the first sign of trouble when just making a purchase and choose another similar title instead. Unless an automatic upgrade should occur then there is no way to save those users or, more importantly, maintain positive word-of-mouth sales from such users. One bad review won't destroy an app's reputation but hundreds most certainly do. Personally I am just looking into adding multiplayer support to my first title and this thread has been an enlightening read. I will, as previously suggested, 'ask them' myself, but since there are members of the 'Exit Games' team posting on this thread are there any replies from them on this particular aspect of this discussion forthcoming?

 

Photon Cloud supports "CCU Burst": https://cloud.exitgames.com/Pricing
You have enough time to upgrade your account (or not).

 

This is what I mean by asking them

Don't bother postulating huge failure scenarios, when they are a very friendly and active team willing to answer your questions.

If you're really concerned, talk to them more - particularly when you've got something worth releasing.