This game will never scale and here is why

[Wollkneul]

Henrik in his recent Stream said, that they expected linear scalability and didn't expect these problems.

But there is a mathematical reason, why it's impossible to scale MO2 or any other game linearly.

In the old Forums I suggested 3 years ago to develop MO2 on Google's Stadia in order to make it linearly scalable.

I was wrong about Stadia. It's a complete mess.

I wasn't wrong about scalability though, which this launch shows:

"The classical bottleneck of every online game is the fact, that your bandwidth requirements with every player and object increase exponentially.

Let´s say you have an online game and each player generates data of 1 mb/s.

With 2 players, each player needs to download data about the other player, so it´s for every player 1 mb/s to download and for the server 2 mb/s to upload.

With 10 players, each player needs to download 9 mb/s and the server needs to upload 90 mb/s

With 100 players, each player needs to download 99 mb/s and the server needs to upload 9900 mb/s"

10 times the players, around 100 times the amount of Data the Server has to handle.

Even if you add NA Servers, you will still have this bottleneck and be capped in the amount of players you can handle.

 

[ArcaneConsular]

its kind of sad they expected their game to flop so hard that they didn't even add a queue system until a week after release. They should've at least had a back up plan or something, like alright lets disable building for the first month and then make x amount of Myrlands that we will slowly merge over the course of 3 months or something. It only annoys me because it was so easily preventable. It's not like 400k people were trying to play. I even posted on the feedback forum for months telling them it would happen but they didn't make the changes. Even simple changes like increasing the spawn rates, adding an afk timer, banning macros, but waited til too late. So now everyone who isn't a NEET leaving in Europe (granted some europe has the largest % of NEETs due to social welfare) can't play for the first two weeks, maybe even more

 

[SilentPony]

"The classical bottleneck of every online game is the fact, that your bandwidth requirements with every player and object increase exponentially.

Let´s say you have an online game and each player generates data of 1 mb/s.

With 2 players, each player needs to download data about the other player, so it´s for every player 1 mb/s to download and for the server 2 mb/s to upload.

With 10 players, each player needs to download 9 mb/s and the server needs to upload 90 mb/s

With 100 players, each player needs to download 99 mb/s and the server needs to upload 9900 mb/s"

10 times the players, around 100 times the amount of Data the Server has to handle.

This is wrong. Players only need local information (about objects around themselves). They don't need to download information about a player #100 on the other side of the continent. So, this relation is wrong. Also, your initial relation is simply quadratic f(x) = x^2, not exponential.

Also, the bottleneck may very well be in the processing power of the server, not in the bandwidth, since it does need to process interactions between all characters. But even that has optimizations that use locality.

 

[Emdash]

Henrik in his recent Stream said, that they expected linear scalability and didn't expect these problems.

But there is a mathematical reason, why it's impossible to scale MO2 or any other game linearly.

In the old Forums I suggested 3 years ago to develop MO2 on Google's Stadia in order to make it linearly scalable.

I was wrong about Stadia. It's a complete mess.

I wasn't wrong about scalability though, which this launch shows:

"The classical bottleneck of every online game is the fact, that your bandwidth requirements with every player and object increase exponentially.

Let´s say you have an online game and each player generates data of 1 mb/s.

With 2 players, each player needs to download data about the other player, so it´s for every player 1 mb/s to download and for the server 2 mb/s to upload.

With 10 players, each player needs to download 9 mb/s and the server needs to upload 90 mb/s

With 100 players, each player needs to download 99 mb/s and the server needs to upload 9900 mb/s"

10 times the players, around 100 times the amount of Data the Server has to handle.

Even if you add NA Servers, you will still have this bottleneck and be capped in the amount of players you can handle.

I appreciate the math. I say just keep adding more servers. There is a capacity per server, so fill it. I find it hard to believe that sub upkeep of like... 2000-3000 people is not enough to support a single server, each time. Shoot, look at most MMOs they have tonz of servers, and we can assume their servers are way more cutting edge than SV's cuz they are mutli-bil companies. haha.

 

[Wollkneul]

This is wrong. Players only need local information (about objects around themselves). They don't need to download information about a player #100 on the other side of the continent. So, this relation is wrong.

Also, the bottleneck may very well be in the processing power of the server, not in the bandwidth, since it does need to process interactions between all characters. But even that has optimizations that use locality.

I am very aware of that, but this makes absolutely no difference for the argument.
Your total player number is in a direct relation to how many players you can handle locally.
When you double the player count and not increase the size of the map (which Star Vault did) you also double the amount of players that are locally next to each other, or potentially next to each other. So you still run into the scaling problem

Thanks for pointing out the term confusion

But why would processing increase quadraticaly?

 

[SilentPony]

Your total player number is in a direct relation to how many players you can handle locally.
When you double the player count and not increase the size of the map (which Star Vault did) you also double the amount of players that are locally next to each other, or potentially next to each other. So you still run into the scaling problem

Don't they have nodes that divide the world into basically separate processing units? They can just increase the number of those if there are local problems.

But why would processing increase quadraticaly?

For the same reason you described, the server needs to check every action against every character, who knows if you may hit someone on the other side of the area with an arrow.

 

[bbihah]

The main problem is that all the nodes we have in the world right now could probably handle the current population just fine. The problem is SV for some reason didn't consider that about 3 cities see 80% of the population in Myrland and nothing was done to rectify the fact that this would be the case. And I think SV expected around 5k players at maximum, not 10k+.

Its more of a problem with game design at that point, they can have 100000 nodes each able to host 1k people but since there is a congestion where people HAVE to congregate the weakest point becomes the max cap. Which i assume is something around 1k people(as per stress test results anyway) per node. So city wise the game can only host 3k people. If the 3 popular cities were an equal split, which they aren't.

Oh I'm glad I dont work at SV, trying to get themselves out of this pickle.

 

[The Khan]

I wasn't wrong about scalability though, which this launch shows:

"The classical bottleneck of every online game is the fact, that your bandwidth requirements with every player and object increase exponentially.

this is only true on servers / worlds where every piece of data gets sent to every other player.

you can mitigate & avoid this problem with node based solutions, like they're using already.

the difficult part is making sure that players don't over-congregate in any particular area to the point where the problem you describe happens on a single node.


worst case scenario though, you just hard cap nodes at a maximum number of players, and do your best to arrange it so that the caps rarely get hit

 

[Jackdstripper]

The biggest single “world” game to date, Planetside 2, can handle a max of 2k players on one single map.

and here is Henrik thinking he can handle tens of thousands of players in his game which is 100 times more complex and detailed than Planetside 2.

 

[bbihah]

The biggest single “world” game to date, Planetside 2, can handle a max of 2k players on one single map.

and here is Henrik thinking he can handle tens of thousands of players in his game which is 100 times more complex and detailed than Planetside 2.

Well we know for a fact that each game node can handle at least 1k people Going to take a wild guess the map has more than 10 nodes. Thats 10k people that COULD play the game in a perfectly designed world. But seeing as 3 of those nodes are congestions that people want/have to go, its far from reality.


Planetside 2's game design doesn't have the same issue. I'm wondering if this issue came completely unawares to SV, because it shouldn't have.

 

[Jackdstripper]

Well we know for a fact that each game node can handle at least 1k people Going to take a wild guess the map has more than 10 nodes. Thats 10k people that COULD play the game in a perfectly designed world. But seeing as 3 of those nodes are congestions that people want/have to go, its far from reality.


Planetside 2's game design doesn't have the same issue. I'm wondering if this issue came completely unawares to SV, because it shouldn't have.

Except that you cannot restrict nodes to a 1k max. What happens when 2k players want to move to the same node? What about when half the server wants to move to the same node for say a siege? Your view is incredibly simplistic. The current world design doesn not allow for 10k. Not even close. As it is painfully aware to everyone waiting in 12 hours queues.

 

[Wollkneul]

And even Star Citizen, which has an enormous funding, is still trying hard to solve this problem after all those years:

As far as I know their current nodes can only handle 50 players.

You can innovate and fine tune as long as you want, but you will always have to deal with quadratic increase.

So far the only solution I could come up with is streaming the image to the users device in order to escape that

 

[bbihah]

Except that you cannot restrict nodes to a 1k max. What happens when 2k players want to move to the same node? What about when half the server wants to move to the same node for say a siege? Your view is incredibly simplistic. The current world design doesn not allow for 10k. Not even close. As it is painfully aware to everyone waiting in 12 hours queues.

Its like your brain literally cuts out every other word you read.
Stop arguing with yourself.

 

[Jackdstripper]

Its like your brain literally cuts out every other word you read.
Stop arguing with yourself.

When you run out of logical arguments, the personal insults begin. Stop spewing nonsense.

 

[bbihah]

When you run out of logical arguments, the personal insults begin. Stop spewing nonsense.

Not until you stop replying to something you clearly didn't read followed by you having an argument by yourself?

 

[Wollkneul]

There is a scaling solution though that seems to work pretty well.

Improbable and Midwinter Entertainment bring more than 4000 Scavengers players together in a single game world

Improbable and Midwinter Entertainment Bring More Than 4000 Scavengers Players Together in a Single Game World

www.improbable.io

Looks pretty impressive

 

[Albanjo Dravae]

Well basically yall speculating with shit u don't know to be fair. Don't have to prove anything to anyone, we all gamers here.

The game does need a higher capacity and it would be interesting if SV can honestly make a statement about the further development dedicated to increasing population. Maybe its too early cuz they are now "apparently" working on adding servers and it remains to be seen "how" they do and i doubt it will be flawless cuz i doubt they had this as a contingency plan and they are just pulling another bandaid off their ass.

 

[Zibnaf]

ive been in many guilds and games mostly MMO's

no game but Acherons call had a smooth launch (1999(and only mmo back then with no zoning and a bigger landmass then MOII today

and i played archage i played with FoE guild on some truly massive fights like on the day for a castles, amount's of people whas just crazy lagy very very like10 FPS but giver there where like 5 major major guilds with their whole playerbase going with their boats to a rather small island .... it was truly epic and must have been well over 5000+ players on a rather small island and it was still semi doable.

so i don't think there is not and will not be any reason for me to doubt thet hendrik and his team will fix this

and apart from Acherons call 1 i never played a MMO that didn't have queue issues with the sheer numbers of login and that's will stick around the starter area and i don't mean haven that is mend a bit to load balance .. once launch is in its first week this like this I've seen in all games(together with the barrens chat ) there where those problems because at the start a lot of people stick to the starters towns levelling and looking for friends and such

once people spread out more load balancing can be tweaked a whole lot more

this game is slower paced with combat and i like it because it feels more like real battles yeah maby not anime style of big boobies....oow wait...this game is or..woman to laugh at i mean super fast paced 10000rps gaitling guns and magic bullets.. so the bandwidth load per character is waay way less then with a game like ... Path of exile and you can only coop with 6 people. and im pretty sure the archeage battle ammount battle of 10k+ on a lets say a castle fight ... well i dont really see that happen unless MOII had 1+ million players even EvE onlike wouldnt handle 100.000 in a single system

instead of the nagging in /help chat... go explore once you go off haven and in the real MOII world spread out build towns, house go trade spread love.....and then start the wars and virtual bloodspilling!!