Playfab - Backend as a service for Unity developers

I haven't seen much mentioned on the forums about this, but has anybody used PlayFab?

It seems like a powerful API with loads of hosted backend services that can make the whole experience of development on the server side of your game a lot easier.

It has everything from data storage (including JSON support), to social account linking and leaderboards, analytics, stats etc, pretty powerful stuff, and I've only been signed up to their newsletter for a month and already they've posted a huge chunk of new features and improvements in that time.

I've started integrating this into my game, and so far the experience has been pretty good.

Has anyone else used it? Not sure about you but I think the Unity world has been needing something like this for a while....

 

I fully integrated it, then I released that if my game even gets slightly popular I was up for some serious cash. However I could have interpreted their pricing scheme wrong, but if I understood it correctly, if I got to 2000 active users/day, I was up for $2/day, $60/ month. Granted I may never reach those numbers.

The positive I took from integrating it however, was it gave me a push into rolling my own. Since I have proper web hosting for my site, I took advantage of the mysql server included. Now I have full control over my leaderboards and can capture other useful information through a few php scripts (it took me about 3 days to roll my own)

Granted my solution doesnt integrate all that other stuff (however integrating with social media/google play etc requires you to provide session tokens - which they dont have API for generating (which I assume means you need to hook up facebook/google play api into your game in addition to their service to make those features useful)

If you only want leaderboards and have access to mysql, rolling your own aint too hard.

On a side note, if you decide to continue on with it, you can support windows phones using the android api, its all just tied in to the system identifier.

 

Thanks for the feedback, I will ask them about the facebook access token, perhaps they can't automatically generate access tokens for your app due to security restrictions or something? Perhaps it is something they are working on.

But what about scalability? Rolling your own custom solution to fit your specific needs is all good and fine, but what happens if you get a massive spike and say 500 000 people all try to sign in at once? Will your server you're hosted on and your database be able to handle that? I'm sure we know as developers the last thing you'd need when your game DOES get a massive exposure spike is for your backend infrastructure to fall over.

I think what they're trying to do here is stop people from re-writing code which is mostly boilerplate stuff, from what I can see it takes little effort to get all the leaderboards and social friends stuff working using their API, instead of spending hours coding stuff that's already been done.

 

I asked them about the token stuff and didnt get a reply... didnt worry me too much, but didnt get them my business either. Its not an access token, its a session token, which I gather means you have to handle the login to facebook.

Scalability is an interesting one... tbh.. not sure. The servers hosting my webpage/mysql are out of my hands, but I assume (dangerous) that they should be able to handle the load. Again, making assumptions here, that if my game really rockets in success that I could look at upgraded hosting plans. Whether that would resolve any issues I might have I cant say, but im sure the guys running webhosting services know more about it than I do and would be willing to help me get it streamlined.

Since I have control over my php scripts, I can tweak the sql queries that are being used if the need arises as well. This also gives me the flexibility to capture the information I want, and at some point, ill be able to make changes to the game by using data retrieved (thinking changing ad providers, tweaking other minor settings). At some point I want to be able to extend my game level through the same means which will be awesome as I wont need to release app updates to offer users extensions to my game (granted the code has to be in place within my game to support this).

If I had 500,000 users right now, it wont cost me a cent. Get the same on playfab and youre up for a $1000/day. Maybe that will be nothing if im creaming it with ad revenue, but on the opposite end of the scale, if i got 2000 daily users I would be up for $2/day, which aint much, however if im making zero profit, thats a big cost.

On initial integration, playfab was awesome, but after using it for a short while there were short comings I didnt like. Not being able to view score boards, delete users, etc.

Another benefit of rolling my own is I can integrate the leaderboards into my own website easily.

eg. http://www.fingerbait.com/hardkour/hardkour_landscape_scores.php

What do they use at MM for their WP leaderboards? I actually intended to ask you this at one point because I was curious.

 

But that's not true, there is no ways your host is serving 500 000 users and your database serving all those requests, and its not costing you a cent. If you're on a cheap hosting plan, its most likely going to fall over. Do you have your own server or a shared hosting solution?
It would be interesting to hear what your host responds to a question you posted to them about traffic/concurrency limits etc. If they didn't have these limitations any Fortune 500 company could just sign up for a $4.99 hosting plan.

Hmm, from what I've read in their API these features are very well supported. Scoreboards = Leaderboards right?
I think in the latest newsletter they mention you can delete users.

They do support this, it's listed at the bottom of their features section. You can get any data out using their RESTful API's.
But of course, yes, you will have more flexibility rolling your own, but with it's own disadvantages.

We use the publishers built in API's and servers which host all their titles, we don't need to roll our own.

 

Well, not costing me a cent isnt correct as you pointed out, there is obviously a yearly hosting cost (which is about $100). Currently using. http://www.ixwebhosting.com/hosting-plans

I couldnt see any restrictions that might affect me. Unlimited bandwidth/diskspace. Looked through the control panel and couldnt see anything that might indicate a restriction on mysql access, so cant confirm. However, like I said, should the need arise to upgrade my services, the cost will be significantly lower than using playfab.

Yes they do have a scoreboard/leaderboard system and yes you can integrate it as you mentioned, but it still took a lot of control out of my hands, which now after rolling my own I have. I can run custom queries against my data and analyse it as I want. I can see daily how many new users I've got, what they score, etc. I can see the user list through their dashboard, and have some control/information on each user when I click on them, but its obviously still a WIP on their end, as some function doesnt exist, or doesnt work.

The real clincher for me however, was that apple wont support purchases outside of their own store system. As soon as I found this out, the benefit of using playfab over creating my own was lost as I would only be using their system to store leaderboards.

 

One other thing to note, their unity sdk wont build for Windows Phone, but removing the code that saves preferences using a binarywriter got me past this problem and didnt seem to affect using the service.

 

And for interest sake, I used this link to get me started.

http://wiki.unity3d.com/index.php?title=Server_Side_Highscores

 

PlayFab seems very cool and it is something we consider for our small indie steam title.

For a free mobile game i am not sure i would do it. The cost could be too much. But for a PC game like our own with paying customers playing it, its really nice (it got an account linked to steam feature).

If you get 5$ yourself for your game per purchase, well then divide that with 0.002/0.006 dollar and you see how many days they have to be active before you lose money!

 

If people are paying for the game, no problem! Its cheap.

 

We decided to use it for our project after some discussion today

 

Interesting, would be good to hear about your experiences on this thread as you integrate it
Perhaps if there is anything lacking we can raise it with their support and get it added in quickly.

The way I see it, is the more people who start using it, will lead to more functionality that we will need, that they might not have yet, and thus more people asking for it, means MOAR functionality at a quicker rate.

 

Well to being with we only need a really simple system so dont think we find missing features for the next couple of months hehe

 

Hi all, James Gwertzman here, I'm the CEO of PlayFab. I'm really glad to hear you're trying us out -- that makes us happy. I've been in the game industry for 15 years, most recently at PopCap, but before PopCap I ran my own indie game studio (Sprout Games) for several years, and so I know what it's like to be small. We had just 3 full-time employees, all co-founders, and out motto back then was "spend no money".

That said, I still think PlayFab is a great deal, especially if you consider all the hidden costs. I would absolutely have used PlayFab back at Sprout if it had existed back then.

The biggest cost savings is for all the up-front engineering work we've already done. You can implement a basic leaderboard with PHP in a few days, but duplicating everything we provide would take you several years worth of work -- and by then we would have already built a ton more features since this is all we do.

We have 12 people who do nothing but focus on building the most complete game operations platform in the world. And 100% of that platform is available to you immediately, without paying anything until your game goes live. And that right there is pretty huge --- it means you can start using all these features from the very first day of prototyping your new game, and focus all your energy on your game.

This matters because designing and building services that scale is really hard. We wrote this blog post about just this topic. A simple PHP script on a $100/year server using mySql can certainly handle some traffic, maybe even 2,000 DAU, but probably not much more than that. The problem with storing a leaderboard in a relational database is that most databases are designed for many more reads than writes --- writes are typically very expensive, because they often need to lock the table. But games are the opposite -- they do many more writes than reads, usually. So even a tiny amount of traffic can quickly thrash your performance and even crash your server.

And users don't like it when your game crashes -- they may never come back. And with the high cost of acquiring customers, if you're paying $2/user to acquire customers (that's a pretty typical cost-to-acquire a new user), then suddenly paying $0.002 per day to keep that user happy with a rock-solid platform starts to seem pretty attractive

Also there's the cost of managing the server... What % of your time is spent reviewing server logs, backing up data, tuning performance, adding more servers as your game scales up, etc. That can quickly become a full-time job once your game is live, and that's time you're not spending on your game itself.

Finally there's the cost of the tools.. that's actually where we are spending most of our time these days, building better tools for actually operating your game once it goes live. The Game Manager that's live on PlayFab today is version 1... version 2 is coming soon and it's a huge improvement. It's got features people are asking for like permissions and roles, full editing of player profiles, full audit logs (so you know what changed in your game), real-time in-game-item catalog editing, and much more.

Building great tools takes time, and if you don't have good tools then when someone calls you up with a customer service problem you're going to have to do a manual SQL query to resolve their issue -- and that's more time you're not spending on your game

Also, it's not true that you can't use our in-game item-purchase system with games on iOS. It's true that you have to use Apple's system for real-money transactions, but that just means you need to sell your bag of 10,000 gold coins for $5 through Apple. Once that transaction goes through, you give us the receipt from Apple, we verify the receipt to make sure it's not a fraud (that's another feature that you'd have to build yourself -- receipt verification), we credit the player's account with the 10,000 gold coins, and then from that point forward all their in-game transactions happen using the PlayFab system.

And with our system, you have full control to manage the in-game items after the game is already live, without having to re-submit anything to Apple. You can add items, remove items, change item properties, change item prices, etc. That sort of live updates are at the heart of live game operations.

Also we have a pretty decent analytics system built-in that doesn't require any additional work to use -- and we make it really easy to generate custom events. Today our built-in reporting tools are pretty basic, but we have a pretty awesome new analytics system coming soon that does all the operations most live games need, such as segmentation, cohorts, funnel analysis, custom queries, and more. And we're thinking about what sort of stock reports we can provide out-of-the-box that actually help you operate your game more efficiently. Knowing your DAU is not actually that helpful. Seeing that your in-game revenue by 10AM is only 20% of where it was at 10AM exactly 7 days ago is much more helpful

Anyway... enough from me. We're really, really passionate about building awesome technologies and tools for all of you, and we're committed to building the best game operations platform. I want to hear from you -- please email us at "devrel@playfab.com" -- I read all mails we get on that alias, and so do most of our engineers.

 

Its nice to hear from the developers. Could you give more details on how you will collaborate with Exit Games?
We will ship our game through steam using both PlayFab and PUN.

Also when will you change the free 1000 DAU to free overall 30K DAU? And why not keep the free daily 1000 DAU? (well i guess most dont go over that so its bad for business).

 

I'm not sure if it will takes years to develop that. You are probably using mongodb or couchbase, unless you consider the years developing the nosql database from scratch and learning c++.

One thing I didn't see was server side scripting. I haven't signed up to explore more.

I'm considering Gamespark, worthy competitor.

http://www.gamesparks.com/

 

Thanks for coming into this thread James, nice to see you here answering questions about your product

They have a server SDK. So in most cases if you want 100% custom functionality, you can still host your own servers and interface with the PlayFab servers, server to server.

https://api.playfab.com/Documentation/Server

 

Hmm gamespakes also seems nice

 

Yes, I'd include Gamesparks as a competitor. One difference is that they're coming at this space from a TV background, whereas we're coming at it from a games background and have deeper ties to the game community. I've talked to several developers who have compared us head-to-head, and at least in the cases that we've heard about, they've picked us.

The server-side scripting is not yet live -- we should be announcing it in early December. Same for our new Game Manager tool, better analytics, and a few other pretty nice new features.

I say it would take years to build, because it's taken us years to build, and we have a pretty amazing back-end team -- and that's not including all the benefit we get from sitting on top of DynamoDB, RedShift, and other such services.

 

Can someone link to an explanation what a dau is? German google is useless in this case as it is a common abbreviation for "stupidest possible user" here.

 

DAU = Daily Active User. Basically a count of how many unique players you have each day.

 

Daily Active User.

So if you have 1000 unique userd playing your game for one day it = 2$.
If you keep it at that each day then you will have to pay 2$x30 per month.

 

Just for fun, here are some current stats about PlayFab -- these are from a presentation I'm pulling together right now for investors. So 3 of the titles we have powered (Super Monday Night Combat, Planetary Annihilation, Loadout) made it into the top 10 on Steam -- and that's some pretty serious traffic.

Number of developer game titles created on PlayFab since it went live 11 weeks ago: 436
Number of those game titles in development actively making PlayFab API calls (100+ API calls): 42
Number of games currently live on PlayFab: 6
Date when UberNet (acquired technology) first went live: Oct 31, 2011
Date when PlayFab, Inc. was established: Jan 1, 2014
Date when the PlayFab platform was opened to the public: Aug 18, 2014
Number of titles powered by PlayFab that reached the top 10 on Steam: 3
Number of peak concurrent users: 31,050
Number of players registered in PlayFab database: 6,627,453
Number of multiplayer game sessions played: 13,570,385
Number of items purchased in-game: 27,483,546
Number of functions in our API: 131
Peak API calls in one day (Aug 17, 2014): 70,258,780

 

Lets hope our game will make that 4!
(It won't we have zero money for marketing hehe).

 

Hi all! It’s great to see the online aspects of games finally getting more of a spotlight. Like James, I’m also a long-time game industry member, just coming up on nineteen years. And nearly half of that has been spent in the online story for games (the latter portion of my time on game teams, followed by seven years in Microsoft’s ATG team, and now with PlayFab). As James said, the team here is entirely focused on online services, so we’re always interested in hearing your questions, feedback, and suggestions.

Since authentication is clearly of interest to many folks here, I wanted to take a moment to fill in the details based upon the discussion above.

First, a side note – JamesLeeNZ, I’m sorry to hear that you didn’t get a response. We make sure to get back to folks as quickly as possible, whether they’re posting to our forums, emailing us, or adding a ticket to our issue tracker, but if you ever have a question which goes unanswered, please be sure to let us know at the devrel@playfab.com email address.

Now, as far as authentication goes, you have a number of options. All accounts created in our service, regardless of the initial registration method, do result in a PlayFab account in the background. If the account creation is done with RegisterPlayFabUser, the account has a username and password – otherwise, it does not to start (all the non-PlayFab routes to login provide the means to create the PlayFab account as part of that sign-in).

We always encourage developers to go with as low-friction of a process as possible to get players into the game. Preferably, no sign-in at all – something you can accomplish on any platform with a unique device identifier, such as iPhone/iPad and Android. For scenarios where you’re using a service like Facebook or Steam, what you pass to us is the authentication token (sometimes referred to as a “ticket”) for that service. There’s no way for us to request that from the service directly, since at that point, we don’t have any proofs of the user’s identify.

When using a device ID of course, it’s important to be aware that the ID can change for the user (replaced device or uninstall/reinstall of the game, for Apple). So if you use that route, we would also recommend incentivizing the user to add another authentication method later, so that they have a valid route to recover their account.

 

Let me just add that when i emailed PlayFab a few days ago, they answered three mails within 45 minutes!

 

Yep, my questions I've had the last few weeks I've had really good prompt responses, no issues there

 

Would still love some more info on the collaboration with Exit Games.

 

There are lots of ways to achieve low latency. I worked for a big data company for a couple of years that had their own proprietary database that was all bitset based and ran in memory. It was massive (over 1 billion records) and would query and return you results instantly. It was a beast to work with and had it's own query syntax, and other than the socket libraries we built around it's API could only be queried via terminal connections (putty).

There are also so new things coming up that should interest people. The latest iteration of SQL Server has a semi-persistent / lazy storage mode. I can't recall offhand what it's called, but it essentially functions like a document database, yet is still relational. Instead of running purely transactional and guaranteeing storing, it just takes your request and says "ok, got it" and pools the data in memory, then stores it leisurely in the background. It creates a potential for data loss, just as any no-SQL database does, but it's extremely fast and will play nicely with a federated database for scalability.

But... the PlayFab guy is correct... building highly scalable systems is very difficult or at least complex. Building a system that allows for scalability isn't terribly complicated, but building it out so it can be scaled real time based on large fluctuations in load, not so much.

 

@BFGames (and anyone else interested), the question of our partnership with ExitGames (Photon) also came up in this thread - here's the reply I just posted there:

http://forum.unity3d.com/threads/in...end-as-service-for-unity.270525/#post-1840417

 

Thanks

 

It's not too hard to create your own system for most basic backend things.

My mindset is very much different, ever since I wrote my first line of code 3+ years ago this has been about learning , and being able to demand a higher salary.

I guess if you're a one man shop , and this is your full time it makes sense .

BUT if you're just tracking leaderboards you can do it with Rails/Node fairly easily, more money in learning Rails/Node anyway...

 

Yes, I have to agree with PlayFab, that's its demanding to create super-scalable systems, but for hosting simple game specific data (leaderboards, playerinfo, stats, gameitems), it's not so hard todo it yourself, if you have some kind of server already running.
You can get 20k req./sec with no problem doing custom solution.

 

Just want to point out that @Meltdown is providing some solid advice, especially in the hosting.. "Unlimited" to a web host actually means "unlimited as long as we believe it is a reasonable amount of traffic for a small website/blog". http://www.ixwebhosting.com/terms-of-service
If your game communicates with the server regularly/constantly in a session, 500k concurrent will cripple it in a heartbeat. Not so much a bandwidth issue because of size, but connections and cpu. And bear in mind that cpu(s) on the box you are on shared with hundreds of other accounts/servers. It may be fine now and for the foreseeable future, but some random featuring on a blog/lp/whatever and site goes down and huge opportunity will be missed. web servers are not game servers.

Not suggesting you go nuts with your hosting now, but having a plan to ramp up in minutes is way to roll. I haven't looked too much in to webplay, but it looks pretty decent. Also amazon has great plans/systems that let you scale automatically. Traffic goes up... new servers are cloned and spun up automatically.

I say this from (frustrating) experience, having a game fail certainly sucks. But it is no where near as soul crushing as having a game go viral/organic and totally and utterly fail to take advantage of it. The time this happened to my team a few years ago it was devastating. We weren't ready, didn't have a plan or the resources to throw at it. It took us a solid week to get it up and stable. By then the tens of millions of concurrent users that broke the servers dropped to a few thousand. We shut the game down a few months later.

 

Is there any news on the Cloud integration front? Can Photon now be used with the matchmaking service?

 

We're integrating the account systems and webhooks first - we'll have more to offer in terms of integration with our Game Manager later on. Photon Server can be used with our matchmaker, with only minor changes - if you're considering that, feel free to drop us a line at devrel@playfab.com, and we can talk you through it.

 

@Brendan Vanous How well does Playfab integrate with custom databases? So if I have a relational database with custom tables etc, what's the easiest way to get them all talking to Playfab's tables?

 

It really depends upon your goals and the specifics of your current implementation. All the user and title data for in-game use (not talking about the logged event data/analytics) is simply Key/Value pairs in PlayFab. So if the question is "how do I get all my existing tables set up in PlayFab, for use in my game", the answer is likely to be a tool which takes your data and encodes it in some manner - possibly encoding your data into the Values in some way which is relevant to your title. Alternately, if you have an external database you're looking to query into, and you have a REST interface that calls into it, you could just write those calls into a Cloud Script, hosted in our service.

 

We dropped using it for our game. Not because its not a good service, but because we dropped all account related stuff for now hehe

However i got a question out of curiosity:

Whats the best way to restrict user account creation? What happens if a game gets pirated by a few thousand people that plays actively? Then you have to pay but you did not get any revenue.

(I know there are steam auth and such, but if you don't want to integrate it into other systems)

 

@BFGames That's interesting - so you're not going to use any kind of leaderboards, configuration data for game balancing, matchmaking, etc.? Or you are, but you're self-hosting them?

For restricting account creation for a title, PlayFab has an activation key system which was used for Planetary Annihilation. This isn't available in the Game Manager as yet, but if it's something your title needs, we'd love to hear from you (since we prioritize based on our community's needs).

 

@Brendan Vanous Actually not. The game has been made over the last two years by a team of friends from the IT University of Copenhagen's game line. And we just had to realize that its hard to do the project we wanted to do with school and now jobs for most people. Personally i work part time for a game company and i am writing my master thesis, so as you can see the time is limited.

So we actually scrapped all the custom made server code we had for handling accounts, ranked games, and so on. Thats why we looked into PlayFab a few months ago as it was easy to use.
But we decided to make the game more of a CV boost than a possible way of making a profit. And therefore just make it as simple as possible.

Our game will be released on steam, so if i need simple stats and achievements i can do that using their SDK.

So its not really a problem with authentication for our game, was just a question out of curiosity for future projects.

 

@PlayFabJames is is possible to build PlayFab implementation for the Windows 8 Store?

 

@BakuJake13 anything is possible, but so far we haven't had too many requests for Windows Store integration so it (along w/ windows phone integration) has a been a relatively low priority for us. What it would take for us to add it would be a determined customer w/ a clear use case and a high-potential game that makes a strong request for the feature -- ideally w/ several other developers also asking for the same thing

Is this a show-stopper request for you? Tell us more about your game!

 

you can fake it using the android API calls, its what I did initially, using the DeviceID functions

you just had to remove the BinaryFormatter code from memory

 

This is also my number one question. My game is a personal hobby development, so it'd be pretty catastrophic on a personal level if pirated versions of my game start costing me money.

If the available Unity SDK just wrap web API calls as is implied by the SDK page, why is it platform dependent?

Another question: I want to get as close to how players are already doing things as possible, so to me it seems like going with the platform-specific solutions (Game Centre, etc.) where possible is going to be best as it's where the users already are. I've only had a quick look at the website, is there any integration between PlayFab and those services?

 

Hi, very interested in PlayFab.
For a game project, I need to find a MBaaS that cover Asia. Is PlayFab service cover Asia?
Thanks.

 

@BFGames @angrypenguin: Yes, we do plan to support activation codes. The way this will work is that there will be an API to activate the game in our service. You (the developer/publisher) will be able to generate those keys in our Game Manager, so that you can distribute them with your title. A player would enter this to activate the game, so that they can login and play on your Title ID. Without activation, login would fail and those users would not count against your DAU count. If that's something you need, please do let us know what your key dates are, so that we can work with you to validate schedule.

Also, @angrypenguin: We are indeed Web API-based, and we're platform agnostic. Our Unity SDK is designed to be compatible with any target platform. The Windows-specific item is that we don't currently provide direct integration for WLID authentication and the Windows Marketplace.

@Robota: Yes, we do support Asia, and we have multiple regions in that area available for server hosting. If you have specific needs for your project, feel free to send us the details privately at devrel@playfab.com.

 

Cool. I'm not in a rush for this, so just knowing that it's planned is neat. To be honest it's something I'd plan to use only as a contingency anyway.

Yep, I've been playing with the SDK a bit and came to that conclusion myself. It's good enough for me, since it covers the most popular platforms well and doesn't exclude anyone.

For what it's worth, thumbs up on what seems to be a well designed product and great service so far. Even if the timing doesn't work out for my current project there's a strong possibility I'll be back in the future.

 

@PlayFabJames sorry for the late reply. The game is called Dogs of Combat. It's currently on the PlayStation Mobile Store, and it uses the IAP API provided by Sony. However, the store is being shut down, so I'm looking to port the game onto the Windows 8.1 Store. I want to keep the IAP feature in the game, so I've been looking at PlayFab. Do you know if I can just use the IAP part of the service? Also, can I use my own GUI and not the GUI included in the package?