Servers

It’s 2024 and I still don’t have a definitive answer from the developers why (I, we) can’t lease a server in Perth, AU, Auckland NZ, or SG Singapore. I’m now getting inquiries for servers in Buenos Aires Argentina.

Can developers please comment on why we cannot lease a community server?
Just in case we need to justify:

  • we are not leasing an authentication server that contains player information (only Gaijin will have authentication and anti-cheat servers)
  • we are leasing a game server where maps and game modes are in rotation
  • In case a site goes down (eg, EU, NA), you have additional redundancies albeit, at higher distance to centres of population.
1 Like

Probably because they don’t want to hand the community all of their proprietary information…

One week passed and barely anyone noticed.

Through the week, several issues from NA & SA servers keep cropping up intermittently; namely:

  • late spawning in Arcade Ground Battles; battles become lopsided as opposing forces are able to control objective points early without opposition from the other team.
  • high latency in Realistic Air Battles then, then normalising to usual latencies and no PLs
  • Game engine run-time parameters handling near simultaneous events (opposing players firing munitions of near equivalence, eg: M109G - M107 vs 2S3M - 3OF25) cancel’s the defeated player’s shot because of desynchronisation; this is serious imbalance.
2 Likes

This is a wrong section to ask devs answer. Only users are here.

I was hoping for a developer to comment.

Otherwise, I would have gone to reddit.

Moderators, please note that this post is being flagged as testament to my inquiry for developers to respond to my inquiry regarding leasing a server that is community funded. This is not a provocative post of any kind.

It may not be provocative but it’s also bordering on the oblivious. The reason(s) you cannot lease a community server are many, so let me put down the most obvious ones:

  • You aren’t leasing “a server”, you’re leasing many servers that form the game server cluster. We’re talking not in hundreds of dollars but thousands, if not ten thousands. You aren’t going to get that money out of the community.
  • If you were considering leasing it yourself in the end, Gaijin would have to trust you with their server software which even if there are NDA’s involved just isn’t going to happen because that’s just normal business, anywhere in the world.

Actually those two cover it quite nicely. And yes, I’m not a developer, but I am capable of using my brain to figure out why this wouldn’t work ;)

Side note, there is an SA server already for Oceania and SE Asia.

  • What do you mean by many servers? I worked in a data centre and 1 physical server has the capacity of providing up to 8 similar services so 1 physical server = 8 equivalent game cluster. How did you know that a game server is composed of a cluster of servers? Do you understand that Gaijin Entertainment has open-sourced the Dagor Game Engine?

  • It’ll be easy if it’ll be just 1 person leasing the service. It’s like a car rental (you lease the car and not own the car). However, after consultation with knowledgeable people here, the leasing party will need to be a registered business operator, so that the accounting and taxation can be properly handled. NDAs will be required for the game, not for the an open source game engine.

And also, SA server is unreliable. They go on and off as if hamsters are spinning the wheels that’s why the servers in TX, USA is better.

Oh my sweet summer child.You heard the bell toll but have no idea where the clock tower is at.

1 physical server can run maybe 8 matches assuming each match has a dedicated core it runs on. Now take a look at the numbers and see that on any given server, there can be over 1000 active matches. 1000 divided by 8 is how many, exactly? Well over 100.

Because I work as a devops engineer for a living? And I know how scaling works? And I’ve been doing this probably for longer than you’ve been alive?

Sure do, I have a fork of it sitting in my own git repo. But it’s just the client engine. There is no server component in there. It’s like a car engine, really. Sitting there all pretty. But you’ll have to build the rest of the car yourself - and Gaijin isn’t going to give it to you because it’s theirs.

So basically at that point you’re just a middle man. Complicating things for no reason at all, because Gaijin can just go straight to the source and keep you out of it.

I think your attention and effort is better focused at convincing Gaijin they need a 3rd server for South America than to try to convince them to let you run one. I mean, props for the idea and bringing it up but lets be realistic?

So you’re dev ops but fail to understand that if Gaijin is the one running the service, they can field as much server as they need. Also… math?

Let’s do the math:

Over 1000 matches divided by 8 cores per server (assuming you’re an Intel boy) = over 125 servers

But it’s 2024… and there are 32 and even 64 core AMD Threadrippers that are like two years old so…
your credibility as dev ops is now a bit sus. And you’ve been doing it since I’ve been alive? HA… How did you get to be dev ops? I definitely can rescue a dog to work for me as help desk.

I really suggest you keep quiet on this and let the developers speak for themselves. As dev ops (LOL), you even failed to mention how authentication services come into the picture, and you think a game server does everything for the every client to handle?

It’s 2024… It’s not 1990 anymore.

Threadrippers are consumer CPU’s - you want to look at the Epyc for server duties. A game server doesn’t necessarily need a lot of horsepower, it needs memory. Minecraft and other procedural generation games being an exception if you don’t pre-generate, but that aside. So let’s go with the AMD EPYC 9754 with 128 actual cores and 256 threads.

So one server can do 128 matches. At how many Gb memory consumption per match? We don’t know that figure, so let’s assume 4Gb. We now need 512Gb of memory which isn’t unheard of in high end machines. So you’d need 10 of those.

But that’s a problem if one keels over. That’s an instant loss of 128 matches and almost 10% of your capacity. Which is why, if you scale things properly, you spread it out more (also known as horizontal scaling as opposed to vertical scaling which is just throwing better hardware at the problem) so you can have resilience and fault tolerance.

And that’s where my numbers came from. Real world data.

Well, then your help desk will suck when all you get is bark bark. As to how I got to be a devops engineer, oh, I don’t know - maybe because I’ve been doing system administration and developing for the last 30 years? And I’m good at it too :D

No shit, sherlock - I left that out because for the purposes of hosting a server it’s not relevant because the auth server will always be central - there is only 1 of them. So maybe you should take your own advice and not speak about things you know nothing about ;)

But the specific detail is, it’s running the server part of the Dagor Game engine. You got specs for that?

Also, how much of the player population in the region will the 128 core and 256 thread of the EPYC 9754? 512Gb of memory for a server? That’s just 4x of the total memory I run on my workhorse. And I’m not even working on textures and rendering. Will you really spend that much to serve less than 15% of the player population? Talk about barking of the things you know. You’re good at it (barking)? When will you learn to scale properly based on economics? 30 years? Add another 60 years to your administration and development and when you learn to balance the books, let my dog know. You better stick to your administration and (so called developing) and leave the operations to Gaijin. I don’t even know why you’re dribbling on

I guess you remembered when Gaijin servers got attacked and no one could log in. Well… I was playing all day long in the holidays and luckily I was already authenticated when it happened. So between you and me, who do you suppose will benefit from knowing how to scale? Wasn’t it you who brought it up? Take note, you’re not only barking at me, you’re telling Gaijin how to run their services too.

I/we don’t need authentication servers. We only need the game server part because people are suffering from high latency and repetitive packet losses from the data carrier routing packets differently when the traffic reaches higher thresholds.

No, and neither do you - but I can make an educated guess.

Neither is the game server. What, you think it sits there rendering everything?

I’ve been reasonably friendly and polite, the fact that you have to resort to this kind of stuff just shows you talk the talk, but do not walk the walk. Learn to scale based on economics? Been doing that for a long time. I’m responsible for a 5 figure budget - it used to be 6 figures, but since I’m kind of good at what I do, that was brought down. CEO is quite happy with me, really, since I make them money.

Don’t need to learn to balance the books either; that’s what our accountants are for.

But hey, tell me more about how you have no idea how things work out there in the real world :)

sigh okay… here comes session management 101. When you start the game, it requires a session. If you don’t have one, a session will be created, for which you need to authenticate. Your username and password get sent to the auth server, which will mint a session key, which is then returned to your game client. The game client then uses the session key to, guess what, play the game.

If you already have a session, you don’t need a new one, hence you can play when the auth server isn’t working.

I’m not telling anyone how to run anything, that’s what you make of it. I just pointed out there is 1 single auth server. Again, you run your mouth about things you don’t know about :)

So why bring that up in your argument? You just contradicted yourself.

Also known as “working exactly like it was meant to”. But please tell me more about routing. I haven’t heard a decent explanation about BGP in ages.

No you can’t. Why? Look at what you said.

Do you even know how the Dagor game engine works? It’s not like Unity or Unreal Engine. Educated guesses? Get some more education.

You mean incoherently stubborn. And bragging about 5 figure budgets that used to be 6? What happened? CEO realised how good you are at “Failed Task Successfully” and decided to scale you down? Alright. You’re good at it. Oh BTW, that 15% figure of the playing population is a puzzle. There’s math behind it, but I’m quite sure you’re good at it.

What’s the matter, not enough brain cells? You think I don’t know how authentication works so you explain your “session management 101”… Why won’t you go back at the start of the thread and read.

Like I said… we don’t need authentication servers and what I/we are asking for is a community funded game server.

I can. Experience ;)

Do you? Fun fact: the client is not the server. The client renders the game, and communicates with the server which does nothing more than keep track of where everyone is, projectiles fired, damage done, and that’s all it does. If you think otherwise, you, my little aggressive friend, need to learn a little bit more about how multiplayer online games actually function.

But I’m sure you won’t be able to explain how the Dagor engine’s netcode works. But prove me wrong :)

Something about pots, and kettles…

Ah! Such wit. Such eloquence. You’re just a troll, aren’t you?

You brought up the point that I did not mention the authentication server in your argument, and now you’re going to tell me oh, yes, we don’t need it so it was irrelevant to bring up in my argument to begin with? Yeah, good on you.

Your style of debate reminds me so much of a toddler.

I brought up the point you failed to comprehend (thus proving "Failed Task Successfully). I ignored your post knowing you’re just another forum troll looking for attention.

I’m not here to indulge your multiple levels of “years of Failed Task Successfully” . The tech trees and events are too much to grind.

Pot. Kettle.

Just you