Third Party Photons non sense pricing for On Premise (Self Hosted)

Hi,
(Rant warning, excuse my hot head)
I am wondering if anyone can explain the completely lame pricing model for Photons On Premise plans.
For starters, I want to make it clear I think Photon Clouds pricing is fair for most part, for what it is.
On Premise, on the other hand, which is where you host everything yourself, i.e. nothing is hosted by Photon, at all. This is very poorly priced, which, I believe is because they either put little to no thought into the pricing, OR priced it solely to make the pricing for Cloud seem more affordable, thus directing the vast majority of sensible users to go with Photon Cloud, instead of going the self hosted route as Photon does not want to compete with other web hosting services.

Now for pricing, obviously these are harsh estimates based on some benchmarks, but for our game to rent a high performance dedicated server that can handle around 1000 CCU safely, will run us around $100 USD/month.
For anything above 500 CCU, Photon On Premise requires you to pay for the unlimited plan, which will run you $175/month per server, which is almost double the cost of renting the actual physical server!

My main points/questions are the following:

1. Why should any self hosted license be based on your player base? Does Photon just assume that because a game has lots of players (high CCU), means that a company is generating massive income and can afford to pay ludicrous pricing, this is non sense and an unfair assumption.
To put this in perspective, Unity's pro license is required only IF your company makes over 200k/year, Imagine if Unity tried to charge you based on how many people were playing your game? This would be a terrible and unfair pricing model.

2. Not only player base, but why on earth should pricing for On Premise be based per server? Every server is vastly different in terms of performance, this makes no sense either.
e.g, if I had 10 low performance servers, this would cost me 10x more in licensing costs than having 1 powerful server handling the exact same amount of users/CCU!

4. What do I propose? I think for On Premise, Photon should move to much more realistic modern pricing model.
Such as the following: (Please keep in mind I am talking about self hosted not Cloud)

Hobby License - This should be a free, non commercial license. If high CCU counts, etc. are detected on a user with a hobby license, Photon should contact that developer and make sure they know it is only for non commercial use, and take whatever action necessary. They have no issue sending me an email every time our CCU count is over our current plan, so I know they frequently monitor this same data anyways, this would be no issue for them.

Indie License - This should be a license for indie developers who make less than 200k / year. The pricing should be anywhere around $30 - $200/month No limits or restrictions at all. And of course as with the previous license if Photon suspects a company is generating more than this, all they have to do is reach out to them. For the most part game developers and companies are honest people, I do not think this would be much of an issue, if at all.

Studio License - This should be for larger studios that make over 200k/year, and the pricing should be anywhere from $150 - $1000/month.

Anyways, I think that is it. Please don't tell "If you don't like it, don't use it!", Photon is the leading multiplayer solution for one of the most popular game engines in the world, I think this is quite important and needs to be addressed.

Sorry if this all came off a bit hostile, I am quite a bit frustrated by this as I am sure you can tell.

Let me know what you guys think,

Cheers,

For reference here are Photons pricing plans:

Photon Cloud - https://www.photonengine.com/en-US/Realtime/Pricing

Photon On Premise - https://www.photonengine.com/en-US/OnPremise/pricing

@tobiass

 

I am sorry you're frustrated.

I can't comment on the pricing of the server-renting business. Maybe it's not extremely difficult to setup a machine, so you can rent it. Why hardware is not more expensive, beats me.

On Photon: You would use Photon OnPremise, when you want to customize the server logic, or if you want specific machines to run the Photon Server. If you need neither, you could easily just use Photon Cloud which runs on shared machines and has a fixed set of features, which works for lots of games.

The pricing is per server, even if you host it, yes. Bigger games have higher costs but we hope that games with more CCUs also make more money. The alternative to this pricing, would be to make the licenses so expensive that selling just a few keeps us afloat. That's also unfair.

One important thing: You put an imaginary cap on our Unlimited License by saying it limits you to 1000 CCUs per machine. That's a much lower CCU count than what we get on mid/low end bare-metal machines and there are two ways to improve at will: Get bigger machines and or optimize the logic.

Halving the price by getting 2000 CCU on one machine is definitely not a miracle.

 

1) $175 per month for unlimited CCU is not an assumption of "massive income". Your average 1st world fast food burger flipper makes that in about 2 days of work. Photon allows you to change your licensing as your needs grow, so if you don't need unlimited CCU yet don't pay for it until you need it.

The intended use for an on premise license is not for monthly rented servers from a 3rd party. They put the intended use in the license's name, so I don't see how they can make it more clear.

2) Instance based licensing has been standard practice in the software industry since the beginning of licensing. Licensing based on hardware performance would be nearly impossible to enforce.

4) Contacting developers regarding licensing is not free, and doing so for customers on free licenses is likely a giant waste of money and a hassle. Why would they want to replace a license enforcement system that is already automated with one that burns countless man hours of labor?

As far as a Hobby license, just because someone is in theory not making money doesn't mean they should get software, which cost a lot of money to make, for free. Limiting CCU count to a low number makes a lot more sense than playing "is this really your hobby?" police.

For an indie license, it is easy for Photon to enforce CCU and per server licensing, while impossible to enforce revenue licensing. Unity gets away with revenue licensing because they have secondary income streams from the Asset Store and other services, and serious developers generally want to remove the Unity splash screen so will buy at least a Plus license. Photon doesn't have any of that.

 

1) You are trying to argue that the main use for On Premise is solely for companies that have data centers around the world ? I can guarantee that 99% of indie companies will not do any hosting, by the proper definition of On Premise. Do you really think that a company that can afford to run data centers around the world, is going to be using a third party networking library...? With this logic, Unity is a "Game Engine" therefore it should only be used for making games. Photon actually lists suggested hosting services in a section of the On Premise documentation.

4) But you are missing the point, Photon already has a free license for people who do not make money using their platform, so in what way is this actually different ? Obviously if an application has a high CCU usage, they probably are not hobbyists, this is not hard to manage or enforce.
Also, as I said already they manually contact every single developer when their Photon Cloud usage is above their current plan, for example if I have 5000 CCU, and I tend to have 5300 CCU, they will be sure to send you an email. So no, it would not be "countless man hours".

Regarding what you say about enforcing the license, this is simply not true, it is very common, and very possible even for very small companies to use the pricing model.
There are dozens, for example, Megascans, and Game Textures, 2 smaller texture/scan companies both offer an affordable license in a similar method, both have license with revenue caps.
Now enforcing something like texture/scan licenses is 1000 x harder than enforcing something like Photon. With texture/scans, they have pretty much have no way of knowing if you have the proper license or not. However it probably does not matter because most game developers are innocent, good people, and are not planning on breaking the law and risking their companies integrity to save a few bucks...
Photon on the other hand has access to all your statistics, data, APP ids, etc. If you think that most companies that are making a good amount of money / year, plan on risking big legal trouble instead of paying a reasonable fee/month for a license, please be sensible.
Also the argument that "it costs them so much to maintain Photon, so you need to pay insane prices for On Premise self hosted" Please.... it is a 10 year old networking library. There are $50 assets on the Asset Store that have had years of development and constant support (eg. anything by Amplify, etc.), and arguably something like that is much harder to maintain with constant Unity updates than something like a networking library. There is absolutely no reason an On Premise license JUST to use the networking library should run someone $175/month/server.

There is no fair pricing or sense behind Photons On Premise pricing, as was clear in @tobiass post.
He was unable to make any fair point as to why the pricing is how it is, or any reasoning behind it, because there is not any.

Also @tobiass

No I did not put an imaginary cap on the unlimited license, yes, it is unlimited however, as I said in my post those number are harsh estimates from our testing, specific to our game, also while on the topic the "unlimited" plan, this is a bit deceptive, because actually almost every server IS limited by specs unless you have some insane super computer that can handle huge numbers, therefore yes, in a sense it is unlimited, but realistically it is actually VERY limited depending on server specs. Some companies might want to have many low performance servers in many different regions, instead of having 1 ludicrous computer just to get around paying Photons flawed pricing.

 

1) I'd expect companies who do not have their own in house servers or datacenter space to be looking at the Photon Cloud product rather than the On Premise product.

Really though, why would Photon want to chase their users to try to figure out revenue? That is a hassle they already figured out how to avoid.

4) Yeah they already have a free license, with limited CCU. The low CCU limit already ensures it is unlikely to impossible to be used in a commercial game without upgrading to a paid license, and without having to employ license enforcers to berate high use individuals or companies with high CCU but claiming to be doing it only as a hobby. The current system already works for them, and allows for free hobby usage. I'm saying removing the CCU or server instance cap for the free license because for some reason you think hobby users need that, is just ridiculous and would cost Photon money to enforce they currently are not having to spend. Why would photon really want to do that? If you need more CCU for your hobby project than the current free tier allows, just pay for the damn software.

 

1) Again, I don't think you understand what their plans are intended for, you are saying that anyone who does not have in house data centers, who wants to have anything server authoritative (which is the main upside for most people of OnPremise) should just suck it up and use Photon Cloud? I am sure this is not Photons intent with these plans.
For any game that does not want cheating, Photon Cloud is simply a joke.. you should not have to obfuscate variables in memory just because you cannot have any server sided logic. The only option for security with Photon Cloud is having every important variable handled by the master client, unfortunately, the whole master client set up is pretty useless and extremely limited. To be fair, older Call Of Duty games were known for using a similar method to Cloud, and were completely over run by insane amounts of cheaters, host migrations, terrible performance, and other annoyances. I can assure you that almost all well done multiplayer games these days have absolutely no important data handled by any clients what soever, this is a very dated approach to multiplayer. Imagine if banks trusted everything clients told them .
The only other option is Photon Cloud Enterprise which I believe only for people with over 5k CCU, which is an extreme minority, this allows for some restricted server sided plugins, as far as I know, which is pretty useless for the majority of people.

4) The "hobby" license is the very least relevant part of this entire thread. For starters, I think you are vastly over estimating the amount of people that would have high CCU usage, it is hard enough to deploy a game, let alone have alone play it, please be realistic...
Anyone that maintains a high CCU usage, I am sure would be happy to pay a fairly priced license.
If for whatever reason Photon was seeing common higher CCU usage in free tier licenses, then obviously it might be smart for them to put a cap on this free tier, if they were unwilling to reach out to the developers, but again you are overestimating how frequent this would occur, trust me.
My main point is that I am trying to make, is for something self hosted, it should never be based on your user base or server count, this is simply not a fair pricing model and there is no logical way to argue in favor of it.

I hope you can see my view point.

Edit: You edited your post as I responded.

So to further reiterate,

"Really though, why would Photon want to chase their users to try to figure out revenue? That is a hassle they already figured out how to avoid."

As I have said in every post, the vast majority of developers/companies are not criminals trying to rip other people off, please be sensible about this. It is a shame you think so harshly about the integrity of other developers with such a bright community such as Unitys. If this pricing model was flawed, why would it be so relevant with smaller successful game developer orientated companies/start ups?
Again, you over estimate the amount of developers reaching these high levels of concurrent users, Photon is not Google, they do not need teams of people to reach out to a few developers that have high CCU and make sure they have the correct plan.

 

You may want to take a look at SmartFoxServer's elastic licenses.

 

@g_a_p Unfortunately our game is already built, however that looks interesting, best of luck with your venture.

We will probably just end up making our own library similar to Photon, api, etc. so it should be painless to move away from Photon completely, and this will save us huge amounts of money in the long run.

It is a shame as I used to recommend Photon to people, I simply cannot after seeing their ludicrous pricing for self hosting.
I will suggest that people to stay clear Photon until they figure out fair pricing for their On Premise plans.
Hopefully this post shed some light for potential customers, I really suggest Photon address this as soon as possible.

 

@deverolirc, thank you for your "good luck"... but actually SmartFoxServer is on the market since 2004, way before Photon Maybe you can take a look at the showcase on our website.

Believe me: venturing in creating your own solution (server + client) will be a HUUUGE work. Be it Photon, SmartFoxServer or else, I strongly recommend to stick with an existing solution which best fits your needs (feature and cost -wise).

Actually having an existing project to test a different multiplayer server is not so bad. You know what you need to implement and you can achieve it easily, once you have a basic understanding of the new platform you are approaching. You can get quite a lot of experience on an actual project, without mixing the game design with learning the new API usage.

 

To host small machines with Photon, get the $95 license for 500 CCU each. Combine at will.

Thanks for saying we're basically not working on our products as they are old. I guess you're happy using only the server without a fitting client library for current-gen Unity. We can send you the old server SDK if you're happy with the features it had and I guess you give your game away after a few weeks, when you're basically done developing it.

Yes, we could charge a percentage of the revenue for a game. It might be fairer or more affordable and we might even make more money in the long run. However, it's simply not what we do now and it would be a mayor break with current pricing, which complicates matters.

To be honest, I didn't care making any points as I had a vague feeling you already made up your mind.
And it's fine.
We just don't agree in this matter.

 

I am not saying you guys do not work on your products, quite obviously you do very frequent updates, as you should, because you are running a business.
You are acting as if this is donationware, Photon seems to clearly be a profitable company.
I am also not saying you guys do not deserve to be paid for your work, of course you do.

What I am trying to say, is there is no logic behind the pricing of On Premise because it's pricing is solely based on assumption. (This assumption being that if a game has many players, they should pay Photon more).
Please keep in mind, I think your pricing for Cloud, makes complete sense and is totally fair for what it is.

So, can you explain to me, and future users who read this, why On Premise, which is a self hosted license, should have an increased premium depending on concurrent users, or server count ? It is important to understand that a game with 100 CCU, could easily, and commonly make more revenue than a game with 10,000 CCU.
With this in mind, it is simply an unfair pricing model, and if you think otherwise I am happy to hear the reasoning!

Edit:

I think you misunderstood one of my previous posts, for On Premise licenses I am not suggesting a revenue cut or %. I am suggesting having a revenue cap for each license, meaning if a company is making over x $/year, they need to pay for a professional license. This is the same pricing model as Unity, GameTextures, Megascans, and many others. They all use something along these lines.

 

Thanks for the clarifications about previous comments.

I and other team members understand what you're trying to say. And to a degree, we agree: A game with 100 CCU may make more revenue than others. In fact, we've seen projects for a handful of CCUs being very well paid. However, it's rare that someone mails us to ask for higher pricing, due to being well funded. We did have caps but if you're not willing (and able) to go into a legal fight about those if needed, you cannot proof or enforce this. We have no idea what revenue someone has. We're not involved in this kind of metric.

why On Premise, which is a self hosted license, should have an increased premium depending on concurrent users, or server count ?​

In our experience, projects with more servers are more demanding (if not technically, then with SLAs and support) and typically also more profitable. If you're running 10+ servers but are not profitable, we can talk about why this is the case. Maybe running more CCUs per machine could help.

 

Yes, with this type of license it is important to realize 99% of developers or companies making decent revenue, are not criminals, and are not going to risk any legal battle of any sort over a small premium for the license, if it is affordable. If you truly had an issue with an individual or company, it is still your service and you could completely close their account if you please.

Yes, that is true, but SLA is only for an enterprise license correct ? This should be a completely separate subscription/package from the licenses I was suggesting.

Anyways, I would really like to see a better pricing model for On Premise in the future, to at least make it affordable compared to Cloud, whether you think this is a conflict of interest, as this would mean more competition with other web hosting services, is another issue, integrity is key.

I am glad I was able to raise some discussion within your team, please consider.

Best

 

We're running more than 2000 CCUs on small machines in the Cloud, with lots of CPU and Memory available. This is why we think the pricing of OnPremise is fair, considering there's a comparatively simple framework to build your own logic and to scale servers.

We do keep your input in mind but please don't expect a change too soon. It's quite some effort to change those things. We've been there.

Thanks.

 

What specs are those machines if you don't mind me asking ?

 

My colleague pointed me to our docs.

This is a standard configuration:

• Quad core CPU (e.g. Intel Xeon E3-1270 v3, 3.5GHz).
• 8 GB RAM.
• 1 GBps NIC / uplink port speed.

More here:
https://doc.photonengine.com/en-us/onpremise/current/performance/performance-tests

 

What free solutions can do the same or better than Photon server? I've searched for a while and couldn't find any.

 

That is probably not a good idea for the most part, unless you are very proficient in networking.
We are starting with Photon Server, and then depending on our user base and how costly Photon is, we will invest in writing our own custom networking library similar to Photon.

 

One thing I want to point out, is that for practically any real production game, you're going to need authoritative custom server logic. For example, you can't even implement server-controlled NPCs with photon cloud.

 

There are lots of production games where one of the players acts as the host, including AAA games. These games though increase the opportunity for cheating, or can provide an unfair advantage to the host player.

 

It's also very common in the indie scene where you want to minimize recurring monthly costs for online support.

I'm a little bit surprised your team didn't look into the pricing before building the entire game to the point of no return. Seems like a huge oversight.

 

This is more of a problem where there's a PVP element to the game, but the lengths that cheaters will go to in an attempt to gain an advantage means that even server authoritative setups can struggle to prevent cheating.

With no PVP element host controlled games are more than feasible.

 

We are a very small team, and not exactly veterans regarding multiplayer solutions, hence using Photon.
So in hindsight obviously this is easy to say, but I agree, it was an oversight.

I actually read the pricing wrong on their website and only realized this later down the road, initially I thought that the cost was $175/month for unlimited everything,(I know it is quite clear that it is per server, so this is my mistake) and the enterprise license which stated "10 servers included" were dedicated servers which were hosted by Photon, and this is why I thought it cost $1,800/month. Pretty silly on my part.

However, at the same time, it is actually possibly the best approach, whether we had considered/understood the pricing or not, as actually having a working product, instead of spending early development resources on a custom networking library, is much more important.
The custom library route is possible for us now (was not when we started), so depending on how the game does, it is likely we will switch over and go this route instead. With any decent size user base with a fair amount of server in different regions, the pricing of On Premise is just way to steep in the long run.

 

I think with almost any multiplayer game, PVP or not, you want as little disruption as possible, just because there is not PVP, do you want people to be able to have fly hacks? However yes, the more competitive/pvp orientated a game is, the more cheaters.

@Limnage said it perfectly
"One thing I want to point out, is that for practically any real production game, you're going to need authoritative custom server logic. For example, you can't even implement server-controlled NPCs with photon cloud.

Yes @Joe-Censored there are a fair amount of these yes, however most of these are notoriously known for being riddled with cheaters. In cases where this is not an issue, either their playerbase is not significant enough for it to be one, or, they are on console, where cheating is less of an issue.

 

lol I remember the old time....
I have buyed my unlimited CCU 99 Dollars in 2013 (self hosted) ...... Now it's 3500 Dollars and limited to 60 Months.

I know that we have to paid for their work, it's normal, but the difference is so big.

 

It's a shame I cannot just buy this library and use it for my game which will be a "simple" 4 player co-op game that does not have any need for anti-cheating. Players will play it together with friends only to have a good time and if they think cheating is good then it is fine too. So all I need is a good networking library and none of the cloud and hosting non-sense. :-( I'd be willing to pay a good amount of money for a mature networking library.

 

@roka That was a time limited special offer for indies back in the days ... and waaaaay below covering cost ;-)

@AndreasO What would be "a good amount of money" to you in such a case?

 

I don't know. Probably more than 50 bucks but also less than 1000 as a one-time payment.
I can see the knowledge required to build such a lib and that it is worth a lot of money for the man hours involved. But what I also cannot see is why it depends on my game's user base how much I should pay. All I want is some multiplayer support in my game where people can host their own game session on their local computers and 3 friends can join that session. All this cloud non-sense is just not required for my game. If we're talking about matchmaking, master server, STUN, etc.... well, fine, I will host it myself or would like to additionally buy THIS service, if I would need it.

One more thing is source code access. I usually don't want to branch off of an author's source code but it helps a lot in debugging problems and in certain situations it is also a must if author is not willing or able to fix my issues. So in case I would pay say 500 bucks for the net library I would also want source code OR have an legal agreement that I would gain source code access if author cannot or will no longer support the library.

Anything else would be a waste of my time and money.

 

I've never seen a game that hits 5300 concurrent users where your monetization is so low that the photon pricing would even be significant. I'd honestly like to know what kind of game that is.

OP are you sure you aren't confusing user with concurrent user? Your posts seem to talk about those as being the same thing.

 

Hi,

Regarding your confusion about CCU vs user, CCU meaning users simultaneously playing. I assume by "user" you mean people who have your game downloaded. No these are not being confused, maybe for a Steam game 5k CCU sounds high, but for free to play, mobile games this is nothing.

For you question about a game with high CCU, while having low revenue. Pretty much any mobile F2P game without god tier monetization, it would be highly significant. Remember with On Premise you are paying for the server you are renting, + Photon charges per server, for any decent ping, and avoiding language issues, you would need at least 10 different server locations.

Using Cloud/Realtime, I have had a game consistently hit very high CCU counts, on several occasions the game made less than the cost of Photon in a given month. The only solution was to remove burst, and cap at a lower CCU because the cost was so high, this results in players not being able to play on occasions. The game was mobile F2P, we had several in app purchases as well as advertisements.
With this said, I have no issue with Cloud/Realtimes pricing.

However, we are talking about Photon On Premise, TL;DR The issue with the On Premise (self hosted) pricing is they charge per server & ccu. Which, to be blunt, is just a S*** pricing model, solely because they want people to use their Cloud/Realtime service instead of On Premise, because if everyone used On Premise, they would be competing more heavily with AWS, Azure, OVH, etc..

Anyways we will be moving away from Photon completely as soon as possible.

 

And just for curiosity, what multiplayer solutions are you considering? (when you move away from photon).

I’m in the same situation, F2P game with low income per user...

 

It is a custom solution, can't go into details now right now as it is not my place to talk about it, but essentially it will just replace Photon without much hassle. I will be sure to give details on it when the time comes.

As a last resort I suggest analyzing your data, find the areas that are costing you lots & generating low revenue, and simply just take your game down from those locations until you have a better solution. Of course this is a not a great idea, but if your networking is costing more than the game is generating, it may be your only option.

Cheers, and sorry to hear you are in this situation, wish I could have been a better help.

 

Thanks for the reply.

I have been investigating the various possibilities and this new service from Amazon looks promising, looks like they made it for this exact situation. I'm considering SmartFoxServer too, the elastic license seems more reasonable for indie developers. Photon Server Self-Hosted would be my choice (because C#), but the price is not viable for little puzzle games.

 

beware with smartfoxserver, ccu = players +npcs

 

What? So every networked object is treated as a CCU? That's crazy.

What determines what is classed as an NPC?

 

in their api you can use objects for npc, there is people in their forum complaining that its ridiculous. but they stand their ground and said its normal.

 

I agree, if I understand you correctly, it really does sound ridiculous.

 

@kiles: If you consider Amazon GameLift, Photon Server is not the equivalent to compare with. Photon Server is for self-hosting. You could check out Photon Cloud, which has a 100 CCU tier, that's extremely affordable with a one-time payment. CCU are monthly max players at the same time.

 

Ditch Photon for your next project: CCU limitations for self-hosted servers make no sense at all.

Use Mirror. It's free and battle-tested.

 

Again: Not entirely the same. You'd have to run a Mirror/Unity instance per match, right?
Are instances managed somehow? Does it spin up more instances as needed?

 

Mirror is (relativley) slow, insecure, restricting and the developers paid assets are highly questionable.

I’ve reported security issues to them which has not been fixed in over a month, multiple ones. Its API is restrictive, you cant even sync a variable to only a set of clients. Its all or nothing.

Mirror is far from perfect. And I don’t know about battle tested.. If you can provide a source (except the ”benchmarks” that make no sense) that would be nice.

There are much better free libs out there.


As for photon, it has its place. Horses for courses.

 

Could you please elaborate on this? What security issues did you report? I'd like to check it out because I switched to Mirror for my project.

 

https://github.com/vis2k/Mirror/issues/721
https://github.com/vis2k/Mirror/issues/720

Neither are fixed. And also, they don't support encryption which is kinda absurd in 2019. Yet I see lots of people using Mirror and sending credentials using it.

 

I think you're being a bit harsh here.

 

What is harsh about it?

I'm not trying to be harsh to anyone, I am trying to speak facts.

1. Relatively slow, it's based on benchmarks of the execution time of common operations such as RPC's. Also in terms of garbage creation, they are really slow.

2. Insecure, two servere (in my opinion) security flaws is something that should be fixed straight away.

3. Paid assets, the developer's previous works such as UMMORPG, I can totally stand behind. They are products he has created and priced. I have no issue with those. But other assets such as the "List server" which is just a message relay at the implementation is not right to charge $30 US for as it's a security flaw in its own sense. I reverse engineered the thing and created a compatible open source implementation. If you ignore my high level of sarcasm in the readme, the points can be found there (https://github.com/TwoTenPvP/TelePatheticListServer).

4. They boast "MMO Scale", and point to benchmarks that make 0 sense. All they tell you is "Hi we once bought a really expensive server, and had x ppl connect and it didn't crash". That's useful to nobody. What's the throughput, whats the tickrate? Noone of that is answered. For all I know the thing ran at 1 frame per minute which means huge latencies.

5. There are much better free libs out there. This is my personal opinion based on the above points and some more (like feature sets, usability etc).

6. Those two security flaws are actually quite serve if you look into what you can do using them. They should be the top priority.

7. Yes, I do believe encryption should be included in networking libs by default. I see it as a responsibility. It's not difficult. I have seen many people send credentials using Mirror, unencrypted. This violates many countries laws.

 

Well I didn't say you were incorrect, just harsh. But for fun I'll respond to your points

1) With it sitting on top of TCP that isn't surprising. Not all games need fast response times though, such as typical MMORPG's. I doubt the default transport would perform well in an FPS for example.

2) I'd agree that security flaws should have high priority. But a month is not a particularly long time when dealing with free support for a free product maintained in people's spare time. I'd cut them some slack.

3) This is really where my "harsh' comment was directed. Picking on the paid assets is not really relevant to discussing the pros/cons of Mirror.

4) More info would certainly be better, and "MMO Scale" itself is a vague claim.

5) "Better" depends entirely on what you would determine is better for your project. If you built a game on Unet, and "better" for you means swapping network API's with as little effort as possible, then Mirror may actually be the "better" choice. Mirror also has the most extensive library of tutorials available of any 3rd party network API, due to being able to leverage most legacy Unet tutorials. There's lots of people, beginners especially, which will see that as better even if the API itself is inferior to others in certain respects.

6) #720 is concerning, but it is really a feature request. Mirror is supposed to mirror Unet, and Unet did not have an authentication system either. I suspect #721 is also a legacy Unet issue as well. See comment #2

7) Encryption is another feature request, and not a part of original Unet, even though I agree that a modern network API should incorporate at least a minimal form of fast encryption as an option. Authentication credentials can be encrypted before sent over the transport though, so I don't see this as any legal obligation to add this feature. Lots of network API's lack built in encryption, so this is not unique to Mirror.

 

1. I am not talking about any of the transport level things. It's a slow performing library in pure CPU and garbage terms. No matter what transport is used, TCP or UDP.

2. A month is a very long time for something used so widley in production. A few weeks I would argue is perfectly reasonable to fix such small things. But nearing 2 months I think is unacceptable.

3. I do think it's very relevant, it shows the core intent of the development of the library.

4. I guess we're on the same page.

5. Fair, but all my points are mainly focued on people looking to choose a brand new API for a new title.

6. It's not a feature request. They claim in documentation that they can do proper connection requests (something my library also claims, but my library actually does it). If they didn't claim it, that would be fine. That would just mean you are misusing the API.

7. And we are pretty much back at my core critique, Mirror is UNET.


So bottom line, if you are at UNET and just need a bug or two fixed, Mirror is worth concidering. But if you are looking for something new or a long term replacement for UNET for a current game, it's horrible.

 

Fair enough

 

This may interest you https://forum.unity.com/threads/unity-gamelift-sample.686341/

 

Do you think the pricing structure for Gamelift is clear? I've had a look at it and I wouldn't be certain how to work out in advance how much I should expect to pay monthly.

But from what I can see it looks like it is likely to be no cheaper than Photon.