Best MBaaS Solutions

Hi guys,

tl;dr What's your preferred mobile backend solution?

I've been shopping around for an all serving mobile backend solution, and thought it'd be a good idea to get the communities experience / advice on what works and what doesn't.

The spec that I have is pretty basic, I need:

• User account creation (when the user logs in an account should be silently created for them, could use their NSUUID for this) after this they can migrate their data to a social network user account.
• Store a user's data
• Match-make for asynchronous gameplay (i.e. you're online, and you play a bot version of another offline player). Will need server side checks for this.

So far in my cart I've got:

Parse (www.parse.com)

Does all the stuff in the list. However, doesn't really seem to handle automatic user creation very well. They have a feature called an anonymous user creation, but their data doesn't get persisted. I could just use the NSUUID and fill in the rest of the fields myself, but seems a bit hacky. Plus, from other devs I've heard their support isn't great, and if you have users outside the US there can be noticeable lag when pulling and pushing data (don't know how noticeable that is).

Google's Mobile Backend Starter (which is a mix different Google solutions - http://bit.ly/1irhUrQ)

All hail Google. This looked pretty good, not as user friendly as Parse, and I'm not even sure if there is a nice Unity plugin / wrapper for it, but nonetheless it's Google and it looks extremely powerful. Obviously, I want to use something that fits my purposes but you at least know 1) What you're getting and 2) They won't go offline. If anyone has any info on if there's a plugin / wrapper for this that would be great. If not I guess I'll have to wrap it myself (fun times).

Kumakore (http://www.kumakore.com/):

Heard it's the Unity's official partner, so that's something. Although looks a little pricy, and not sure it's as powerful as the other two.

Gamesparks (http://www.gamesparks.com/):

Doing some reading on this one now. Again seems pricing, and seems like it has a really low user cap for the first tier.

Ah well will update if I get any more info myself. Any advice would be great. Thanks!

 

I did some research on this about 2 months ago. There are a ton of MBaas. I settled it down to 3: Parse, Kii, and App42. Each has their pros and cons, but I felt these 3 had the best documentation for working with Unity specifically and feature sets.

These are my notes roughly for the major players:
1. Photon Turn Based - still in early access. Can't release a production game on a beta.
2. App42 - many features (users, social, leaderboards, cloud storage, etc.) Cheaper than Parse but same burst limit as Parse (40api calls / sec)
3. PlayerIO - now run by Yahoo GameServices. They want a ridiculously high % of your revenue. Not even a consideration
4. Google GamePlay Servies - cross platform, has a unity plugin, similar services to iOS GameCenter TurnBased, but they, for obvious reason, don't support Facebook
5. Parse - more expensive end but many features we wanted, good Unity Documentation (but you have to hunt for some of it because they don't link it on their site), great browser view of your backend. However, they have same low burst limit like app42.
6. GameSparks - they use MAU instead of api calls / sec. Good feature set. However, they are way too expensive. We took the numbers from one of our more successful games and it was WAY higher than even Parse. when I say WAY I mean like greater than a factor of 10. Research stopped there.
7. Kii - good feature set, good Unity documentation (they are a Unity partner), a real competitor with Parse. Their biggest feature is that they have amazing burst limit of 150 api calls / sec. That is HUGE. They also have cloud code "server extensions" you can run (which you are going to need if your game is competitive at all) However, I don't think they can do relation queries. Pricing is about the same as Parse. Also they don't have anonymous users (all users must register or you do it for them under the hood). Total picture is that they are near the top of the pack.
8. Kinvey - no unity support.
9. Appcelerator - no unity support.
10. Applicasa - some unity samples, heavily weighted toward a virtual store. Their cost is Average DAU/month. Based on the numbers of pretty popular game, you will end up around the range of Parse (maybe a bit more).
11. Open Kit - just social and challenges, good integration with GameCenter and FB. However, they do not have cloud storage.
12. Azure mobile - mess of different platforms.

Again, these are my notes. They maybe of use to you. If someone reading this (company shill, etc.) gets butt-hurt about any inaccuracies in my notes, post a correction below. But personally I really don't care.

We ended up going with Parse. Truth be told they have had some server issues lately (e.g. All FB users were locked out of Parse for a couple of days. Predictably, this cause some apps already in production VERY big headaches.) Also all their unity documentation/samples are not linked on website (you will have to google for it). But we are hoping they can iron out the kinks. If not we would probably switch Kii or App42.

Hope you find this helpful.

 

My only concern with Parse is that with an app with a European user base, I've heard a few other devs complain about the lag when making data requests outside of the US (I assume they only have servers in the US). Any experience with laggy data requests (that is if you're outside the US) ?

Also I suppose the benefit of Parse is that since it's FB backed it's unlikely (I suppose some have) to be canned. Can't imagine anything worse than having an awesome BaaS only for it to be shelved and you're up **** creek without a paddle!

Atm it looks like a bit of split decision between Parse and Kii. Just checking out App42 now.

Also one thing I'm really looking for is a 2 step auth, as in the user gets an account automatically (silently) created on the app startup and then can link their account to FB later on. I know Parse has anonymous users but the data isn't saved between sessions. I think for most BaaS (Parse and Kii) you can just populate the user details yourself and migrate them to FB (or another social network) later on. If you (or anyone else) has an details re this setup that would extremely helpful.

 

Re Parse anonymous users, the data is still stored as long as that user session is not killed. Parse caches the user session (also the FB login as well) to your disk. So you should be okay if as long as: A) you don't sign them out or B) they don't uninstall the app.

You are correct the other route for a frictionless user experience (and that is what you want in this day and age) is to give them a user ID/Pwd under the hood (make their ID the same as you get form System.GUID and then randomize a password for them). If you want to let them take it to other devices, you can save it to a cloud system like iCloud. You could do it like this guy https://www.parse.com/questions/icloud-and-anonymous-parse-users

Or you could just tell your user that "if you want to play this on multiple devices, you need to link to a social account e.g. Facebook." But if they did, then they could play it on any platform that supports Facebook - iOS, all the flavors of Android, Facebook Canvas, etc.

If you want to go the route of Clash of Clans, you could also go the route of saving this generated ID/Pwd to their keyChain (iOS). That way if they uninstall the app, then later change their mind and reinstall, you could ask them "Hey are you XXXX?" and restore their data.

 

Yeah that's the route I was thinking of.

I think I've settle on Kii, just for the burst rate and price scaling seems a better fit over Parse. One thing that worries me about Parse is the lack of support on their forums, there seems to be a million and one unanswered (0 comments) questions. I mean the docs are pretty good, but nonetheless it's nice to know that if you want some custom cloud code / custom setups you can just quickly discuss it with other people (just like this lol).

Just out of curiosity what made you choose Parse over say Kii? You mentioned functionality, but it seems they have almost identical functionality

 

To the best of my understanding, Kii can't do relational searches. I have limited back end design experience. While I had some idea what the backend data model would be, I wasn't completely sure and didn't want to unwittingly fence myself in. Faced between not being able to do something that I want to data-wise (hence favoring Parse) vs throwing money at a problem (Kii has a better burst rate so you could end up paying less for an app heavily used in just 1 time zone, which is what we are planning), I chose to throw money at the problem (if needed).

Also I felt the documentation for Parse was just a bit better. While it is true there are not that many questions being answered in their forums, Parse has a ton of users. This means better StackOverflow answers in the long run (there is a reason FB engineers actively monitor StackOverflow).

Another thing that I forgot to mention in my notes above is that Kii has a very large Chinese presence. So if you plan to roll your app out to that market (which we are not), you might consider Kii a better choice.

Something that I have not actually tested, but would advise you do after you get some experience with schema-less object saving, is to compare the browsers of Parse vs. Kii. Parse's browser is very easy to use and intuitive. I find that a big help when iteratively testing my code.

Also I found this to be helpful. http://blog.steamshift.com/blog/2013/03/18/working-with-parse-dot-com/#.U6wzYF6fv0B Granted I found it after I chose Parse.

Bottom line: I think they are close competitors, near the top of the MBaas. Both will probably meet most people's needs. I think it will take time and significant experience to draw distinctions between them.

 

I finally (after some extensive research) settled on Parse. Kii was amazing; I really like the community they have there, but after nitpicking through all of our requirements, Parse won by a nose.

It really came down to setting up a native account and then linking it to a set of social accounts. Parse does this seamlessly (being owned by FB), while Kii creates another account in addition to your native account and you then have to migrate the details between them. This only applies to their Unity SDK, which will no doubt will be updated soon with the feature (their other SDKs i.e. iOS and Android already support this).

I've tried both data browsers (Parse and Kii) and there's not really much to choose between them, they both act and behave in a similar way, and there's no clear winner, at least for me.

Your bottom line sums it up perfectly, close competition in the BaaS space can only be beneficial for the customers, with no clear winners, it'll be up to (as it typically is) your own personal preference.

 

Just clarify something I mentioned earlier about Parse anonymous users (in case someone is doing research and stumbles across this page). As of the writing of this post, Parse Unity SDK doesn't support anonymous users. However, if you desire that functionality you can use another one of their SDK along side the Unity SDK. While I haven't tested this out, from looking at the the flow it looks like the way you would do it using their REST api. https://www.parse.com/questions/anonymous-users-with-unity

 

Hi,

I'm doing a research on Mobile Backend Solution, that's why I'm come to this post. I am wondering is there any Solution that did support Anonymous User login and then can be linked with Login Account? This is what I'm always finding currently.

 

Hi AppxploreF,

As Texaggie96 said anonymous users on Parse is not yet supported, however a few people have written code for the REST API and used it in Unity, see: https://www.parse.com/questions/anonymous-users-with-unity

Or if you go for Kii, I'm pretty sure it supports anonymous users out of the box, see: http://documentation.kii.com/en/samples/Gamecloud-Unity/

 

Hi poolts,

Thank you for your suggestion, but I had think of a way by using the CloudCode to create Anonymous Users in Unity.
If anyone is interest welcome to refer to the link below:
https://groups.google.com/forum/#!topic/parse-developers/hWsq1RG_4YE

 

Oh ya by the way, is any MBaaS above provide offline data sync?

 

@AppxploreF Sorry for the delay; I am still recovering from Unite. I am confused by your question. Could you be more specific about what you mean by "offline"?

 

@Texaggie96 Sorry for delay, I'm not login until now. Anyway, I had decided to go for Parse.
I'm now still in the middle of doing it but already have some experience to share. Parse only provide a google groups and stack overflow for question and answer, totally have no support email as i try to search out. That's the biggest problem i had and it makes the developement time a bit longer than i thought. By the way, does anyone know to whom i should email to if there is a question regarding Parse technically?

 

Folks, hope you don't mind me butting in here but I work for GameSparks.

I am naturally concerned that you think our pricing is 10 times more expensive - would you be able to provide some metrics so that I could compare like for like? We go out of our way to make our platform (which is the most advanced for games) affordable for Indies. While a lot of our customers are big publishers, we also work with many indies and generally our pricing is working out for those we work with.

If you contact me directly we will work with you on commercials. We also provide direct support to all of the developers we work with.

 

Hi Johngriffin, My quick, back of the napkin calculations went like this: In the month I did the research, we had 315k MAU (over 4 SKUs for one of our mobile apps). Subtract the 10k you give for free. Drops the price number down to 305K. 305K MAU * 20$/1000 users = $6100/month. While I don't have req/s data for the same time period (most of your competitors use this metric), this cost was significantly higher than others in the MBaaS market at the time. If you go to parse.com and drag the slider all the way to the right to get 200 req/s, you still only get $1700/month. (The published prices at the time for Kii.com were comparable.) Even if your "bespoke" solution was 50% a discount, that would still be, more than likely, higher than parse. To get down to around $1700/month (just for dicussion purposes), would your bespoke solution be greater 72% savings off your list price?

 

Thanks - really appreciate you coming back to me on this. There are 3 main points to make:

1) In general terms, its difficult to compare MAU pricing with API calls / Bandwidth consumption / Storage etc as that varies for each and every game. We have taken the approach of pricing by MAU as it is both simple and takes on some of the scalability risk with for the developer.

2) Comparing us with Parse it not always a like for like comparison. We have a highly tuned platform for games and are designed to be able to, for example, in the case of leaderboards, return real time rankings while millions of concurrent writes are going on. Our platform is specifically orientated around games and is both designed to save you lots of build time with a library of functionality that Parse just doesn't have as well as tuned to the kinds of things games need to do that a non-game app might not. The support side of things is also quite different - we are more actively engaged with each of our clients in helping them get the most out of the platform.

3) Lastly, we do offer great discounts to many of our customers so with a little commitment from both sides we can get well below our list price. Please contact info@gamesparks.com to initiate a dialogue around that.

Once again - thanks for taking the time to respond on this. We are really committed to being the best solution for developers out there and pricing is an important part of that.

 

I just watched the Unite 2013 video on Facebook integration and Parse.

I was pretty surprised by the mixed reviews for Parse when I checked out the SDK on the asset store.

This thread is an awesome follow up. This info is really great. Thanks everyone for sharing your thoughts on this issue.

 

If you want to pm me I'll be happy to see if I can help you with any Parse stuff. I agree there single biggest problem is the lack of an active community / support with regards to Unity. Although, it is definitely getting better!!

No worries It's always good to share information among devs. Especially, when it's months worth of research condense into a single thread.

 

Well, I'm with PlayFab, so I'm going to be a bit biased, but my take on this is the same as John's (from Gamesparks). Companies which are dedicated to building systems specifically for games, and who understand the problems game developers face, are the ones who are going to provide you with the best tools. It's certainly possible to build your tech on a very generic stack, but the downsides are numerous (having to build all the custom logic yourself, do all the monitoring and debugging work for that logic, etc.).

But for me, the best reason is the same one that brought you to Unity to start. Why would you build an engine from scratch when there's already one which does what you need? And which has a team dedicated to making it better, year after year? Your time is better spent tuning your game for the fun.

All the services like ours are intentionally designed to be free to build on (and as competitive as possible post-launch). At the very least, using our services to prove out your ideas in a development sandbox provides you with an easy "leg up" to fleshing out your ideas. If you decide to build out your own services prior to launch and switching over to them, there's nothing preventing you from doing that.

 

Oh ! Well discussion here...
Does parse has leaderboard and achievements built-in?
I don't know why azure is not here, what's the problem with azure? (I think azure has all the things of parse, plus actively development of last 3years(stability) as I know and also an open-source active unity client: https://github.com/bitrave/azure-mobile-services-for-unity3d); am I wrong with azure?

 

Azure Mobile Service uses SQL db by default, so most game dev may prefer to use a NoSQql db for unsturctured data support. They have recently released a new DocumentDB service which can be used with their Mobile Service, but may require a little work: http://giventocode.com/how-to-use-azure-documentdb-with-azure-mobile-services

Personally I do like the fact that Azure support multiple regions like Europe and Asia (also China if you sign up for it), unlike Parse which is purely US based. But for Unity integration I ended up going with Kii because it's slightly easier to setup and also has more Unity related documentation/samples.

 

Yeah Kii was my very close second choice (to Parse). Kii has a burst rate for 150 requests /s which compared to Parse's 30 requests /s (we went with Parse due to it's strong support for Facebook account linking). Mahdi, there are a few backends that are specifically designed for games i.e. App42, but typically you can achieve the same things but with a bit of dev work. It all depends how cost vs dev time and requirements.

 

Bear in mind that you also should plan for the monitoring and maintenance for those services through the lifetime of the game, and factor that into your ongoing costs. Also, think through your sunsetting plan, so that when you want to start moving off supporting the game, you can do so without making your community feel like you're abandoning them. Some questions you should be asking if you want to own the backend on a general hosting service are (there are more, but this is a start):

What's the historical uptime for the service? What do I do if there are outages? What is the service promising (SLA)?
If the service goes down at 2am, how am I (and whoever else is in my monitoring team, since I don't want to be on the hook 24/7/365) notified and how quickly?
How quickly does the service scale and how much work do I need to do (if any)? If I get featured in Penny Arcade and my user base goes 100x, what's that look like?
When my users find exploits, how quickly can I get an update in place?
What's the roll-back look like, if there are problems with an update to my server-side logic?
If I want to move to another service, how easily can I get my data, and in what format?

Brendan

 

I thought this might help people out.
I started a comparison matrix to compare what is out there for MBaaS.

Feel free to make changes if you see any are necessary.

https://docs.google.com/spreadsheets/d/1rWLgMouf8-3zC6gqvGzgLUu3F4zMgZ351Y1eN_ZKFa8/edit?usp=sharing

It is a first draft.

 

This is awesome. Thank you!

 

@UGJackal2 you left off PlayFab... give us write access and we're happy to fill it in
We've got a pretty large team dedicated to building the best platform-as-a-service for gaming - with a particular emphasis on the live ops tools themselves, not just the backend services. We've been operating games for more than 3 years, including 3 games in the steam top 10, and we're fully cross platform (our first PS4 game, for example, launches tomorrow).

 

All done folks. Anyone can edit the comparison spreadsheet now. It will be interesting to see what I missed and/or messed up.

Thanks for the feedback.

 

Leaning towards Parse over Kii (with nagging concern over latency issues for European users). I could not find any documentation on the Kii site for sub-classing similar to what can be achieved via Parse (see here). Also as Texaggie96 stated above...relational querying is a MUST.

 

I've been using Parse for about 6 months now and I'm pretty happy with choosing it. The only thing that I would say is a huge detriment to Parse is their total lack of support. They're trying to get the community to help each other (as opposed to direct support from them) but it's not really happening, and load of questions are going unanswered. Apart from this issue though (and I guess we are getting a lot for free) Parse is pretty awesome.

 

I have been using parse for some simple stuff since July 2014 and I have been pretty happy, until I tried to export the project to windows phone. Apparently, I cannot make it work and I could not find any help online. One thread was abandoned a while ago and people seem to have built their own REST wrapper to parse services which is kind of disappointing.

 

We tried Parse earlier, and as the project gets bigger, we thought its better to reconsider choosing parse because it became difficult to manage.

Then we shifted to GameSparks, which is pretty good in-terms of being game development friendly.
We are still working with GameSparks apart from its minor drawbacks.
Mainly, the forum is quite slow in responding back. It would be great if they improve in the service area. Rest of the stuff is pretty good.

Will update soon if we see any major issues.

 

I am using App42 and pretty happy with support and services they provide.

 

Does anyone have experience with brainCloud? The API seems great and satisfy any need we have for a backend, but I'm not sure about response times or how well it handles traffic.

 

@danlee0710 Thanks for pointing another service.
I see its pricing based on API calls limit. How usually you calculate api calls per user when game is live? I see it always tough to find out that part

 

@PlayFabJames & @Brendan Vanous We have started using PlayFab and are really pleased so far with the support and the performance. It also has some excellent features. A definite contender.

 

Awesome - thanks for the feedback, @SketchWork!

 

We have used Braincloud for a couple of projects and really liked it. In fact, right now we are developing an app that will possibly reach 100K users at peaks, and we're using Braincloud for the back-end. They have great customer support and so far, they have provided us with all the features we have requested. It's really worth giving it a try. They're pretty solid security-wise too (which was an important requirement for our current client). Highly recommend!

 

Has anyone tried AWS Mobile SDK for unity? I think amazon is more reliable because it's the largest web service out there. I haven't tried it, but i am considering using AWS or Parse.. and i'm really confused

Here's the link for more information:
https://aws.amazon.com/blogs/aws/now-available-aws-mobile-sdk-for-unity/

 

I've tried both AWS and Parse. Out of them AWS is bit incomplete and doesn't work on some platforms. And they don't have proper docs or example for a starter.

About Parse, It got a great management dashboard and APIs. And the docs makes it easier to use. However, If you're a starter with Parse you'll find it hard for managing callbacks and you'll experience some weird threading issues. There are no proper documentation for how to overcome such issues. If you're planning using parse, be prepared to google a lot.

The best solution I could suggest for you is Kii cloud. It really works well with Unity and the APIs are easy to understand.

 

Hi, @appslabs

Thanks for the response and suggestion, i think i'll try using Kii because i'm not a really great coder so i want an easier one

 

Hello devs.

I've read all this thread that helped me getting some clear idea on the MBaaS argument. Clear, but not clear enough...

I just posted a question in another thread, so don't blame me if I link it for help, as soon as there might be people that follow this thread and could answer me. Thanks a lot. This is it: http://forum.unity3d.com/threads/mbaas-perfect-service-for-my-game.355397/
Thanks to anyone that would like to help me.

 

Parse is closing down 1 year from now...
http://bits.blogs.nytimes.com/2016/...arse-its-platform-for-mobile-developers/?_r=0

 

Here is a list of alternatives to Parse https://github.com/relatedcode/ParseAlternatives

 

Hello,
this is just to make you aware of kii price levels. As you may know, they don't show their prices.
We asked them a quote of the correct plan able to manage the custom leaderboard of our game (Real Drift) which needs approximately 2 Million API Calls/month, 10 GB Storage, 10 GB/month Bandwidth, and they replied (after some insistence...):

Our first tier of service is 400 dollars USD a month and you get the below. Please let me know if you have further questions or need help with your evaluation.
Our first tier pricing is set out below, along with the overage supplements.

Monthly Subscription: $400
API calls (Million): 10
Overage/Million: $40.00
Push (Million): 5
Overage/Million: $20.00
Storage (GB): 20
Overage/GB: $0.05​

This is OVER 4 times a comparable plan of, say, app42 (http://api.shephertz.com/pricing.php).
I replied asking why this huge price difference, i dindt receive any reply.
Take your own conclusion. To me, it really seems that they dont show their prices cause if you start with the free plan and then you need more than that, you find yourself forced to pay over 4 times the market price (or move your data to another backend, but it doesnt seem to be export functionality).

 

Thank you for this info.

 

Anyone that knows about loopback.io? a open source mbaas for nodejs. They have support for iOS, android and windows not thorough unity3d tho but one might be able to hack it to work either way!?

 

@Darkoo I don't think Loopback is a backend service. It looks like a web server framework for NodeJS. It seems inspired by Ruby on Rails with "generators" which will set up various domain models for your project with database-ready entities. You'll need to deploy the code somewhere and manage all the usual difficulties with logs, backups, maintenance, upgrades, etc.

A clever element of the framework is it integrates a database within the client libraries which is used for remote-sync. This can be useful in certain kinds of offline-first games but depends on how you want to handle bad players who'll manipulate the data in their devices.