Howdy, Stranger!
It looks like you're new here. If you want to get involved, click one of these buttons!
Quick Links
The one server concept
Jackdog
Member UncommonPosts: 6,321
Will it work? The Wish team could not get it perfected, even using PnC to minimize the traffic to and from the server they had lag they could not conquer. Blizzard and SoE get lag on their servers once a certain number of people are logged in and even with Blizzards deep pockets for server tweaking they cannot seem to fix it.
What is so special about this DnL that is so special that it will work there? I am not saying it is impossible, just that it is highly unlikely in my opinion and experiance. Time will tell I suppose if it ever releases. Of course it may go the way of Wish if it looks like the technology just is not there yet for a one server MMO.
I miss DAoC
Comments
First of all - DnL doesn't use ONE server, nor did Wish. - They use a large cluster of servers.
Games like Wow and the like who only use one server for each world have a very limited population on them 2-3k up to perhaps 10k for the technically advanced games / servers.
In DnL there will be hundreds of servers connected to house every player in the whole world in one world. If an area becomes crowded, the client will automatically adjust settings to avoid lag. I think other servers will also kick in to help take load off a crowded cell. This kind of technique would also not require the shut down of the world for maintainance. - Another server will take over the job of the server being maintained. Brilliant, isn't it?
Nemo sine vitio est
Server technology behind DnL - http://www.bigworldtech.com
What about Eve?
Eve does just fine with 13k+ players on at one time. It has over 50k subscriptions, and only one shard (universe). They estimate they can probably host up to around 25k players at the same time.
Assuming DnL has a similar level of competence, they should be able to handle *at least* 20k players at once, which is enough for about 80k total subscriptions. That's probably very near the upper limit for single-shard games with current technology, though.
The limitation is really not the raw number of players on at once... You can always add more physical machines which run more land areas, allowed more players. IF they are spread out evenly. The problem is that players tend to congregate. And the more players you have in a given area, the more stress on the servers, and the more bandwidth required.
The more players on a single server, the more often players will congregate, and the more players will be in those gathering points during peak moments. So with a 2000-player shard, you might see a few random clumps of 20-50, and one or two big clumps of 50-100. But with 10,000 players on at once, there will be a tendency to see many clusters of 50-100, and some clusters with many more. As you add more players to the system, the tendency to cluster creates more and more large groups, and the largest groups continue to grow larger and larger. At some point, the number of players on a server is so high that the size of the largest groups is enough to make the game unplayable, or crash the server. This can be combatted to a degree by adding more servers, and adding much more physical terrain, as you add more players. Giving players room to spread out (and reason to do so) helps a lot.
In the final analysis, though, you simply are not going to see single shard games with many hundreds of thousands of subscribers anytime soon. It's not possible, yet anyway.
Owyn
Commander, Defenders of Order
http://www.defendersoforder.com
A couple of points here, my mistake saying one server. I meant one world, or one shard if you will. Excuse me on that one.
Eve is a whole different animal Owyn. Basically all you have in Eve are a lot of "sky" textures. Add in the occasional space station and asteroid cluster and gate. Not a whole lot of polygons there. Then go to even one of the older MMO's such as DAoC, AO or EQ and take a look around. A couple of trees have more poly's than a whole sector in Eve throw in a few mobs and the polys start to add up real fast. Even the ships are low poly count compared to a character so using Eve as a example is like comparing apples and oranges.
Now back to my question..I do understand what Wish tried to do and how it tried to do it. It failed and failed miserably. Part of that was due to the fact that the majority of users hated the PnC. There were other problems also but trying to jam everyone into one server (farm) was the main culprit behind Wish's failure. This Big World technology sounds interesting but has it ever been actually used in a game yet. Nice theory here but what games have used it before ? Quite frankly if it does not actually work in practice then the game is destined for the cutting room floor.
Since the game is 'unofficially delayed' now and since it happened right after the latest wave of beta testers was let into the game could it be that the Big World technology was a nice theory but that was about it ?
I miss DAoC
Well thats speculation, and saying it for the sake of saying it doesnt mean much... if it isnt simply speculation, then what do you know?
Well thats speculation, and saying it for the sake of saying it doesnt mean much... if it isnt simply speculation, what do you know?
Yep it is speculation, just as you are speculating when you believe that the Big World , one server farm will actually work. However I can put two and two together and it seems likely that this is the case. I mean think about it game is scheduled for release and open beta then all of a sudden they add a stress to the servers and the delays set in. Does not take a rocket scientist to figure that one out.
I speculate about a lot of games, sometimes I am right other times I am wrong. I was one of the ones who actually thought Wish would succeed and Dragon Empires would be on the shelf by now.
I also predicted I would be bored to tears by WoW and EQII and I was. At the moment I am anticipating in no particular order 3 or 4 other games one of which is D&L. I have been a registered member of their site for about a year or more. Lately though I am beginning to have my doubts about whether DnL will be the game for me even if it makes it to release, which I am also beginning to doubt. Time will tell.
I miss DAoC
Actually I never said it would work. What do we know about anything untill we see it for ourselves.
I dont think speculating is wrong... so enjoy yourself
But facts would be better!
Not really, because polycount is only a small part of the problem. Sure, more polys or bigger texture counts make a scene harder for the client to render. But a lot of the issue with large pop servers is serverside, not clientside; or in the bandwidth issue.
Servers don't care about polys or textures, because they don't render anything. They simply process take input, process data, and give output. A game could have players flying one-pixel ships in a big empty space though, and if you add enough players in a small area, the servers will eventually have trouble tracking them all, creating server-side lag. Each additional player in a small region increases the bandwidth and server resources used by a very large amount.
Bandwidth is a problem too. It's not just servers that chew through more bandwidth when there are lots of players around. Clients need to receive that data from the servers, too. This causes problems for 56k connections much sooner, obviously, but put enough players in a small area, and even a strong home-broadband connection will start lagging.
The key to making large pop servers work is in keeping the players apart from each other. Eve does this through design - it has a very, very large universe. And even within the single star systems, there are many specific regions. This lets the server farm easily split up a star system and load share the clients who are there. Of course, if you get enough players into one spot in Eve, you get the same server issues as any other game - but they have made attempts at least to encourage players to spread out through size and to some degree gameplay. The key question should be, what has DnL done to keep players from congregating, and encourage them to spread out? Guess we'll have to wait and see.
Owyn
Commander, Defenders of Order
http://www.defendersoforder.com
I cant wait to see if this works out. I do have one issue with this. With only 1 server will it be possible for one guild (clan) to take over some inportant spawn that sets rest of the guilds way back for years to come.
I just want to point out that expecting boredom and dissapointment usually results in exactly that. Self-forfilling prophecy so to speak.
I just want to point out that expecting boredom and dissapointment usually results in exactly that. Self-forfilling prophecy so to speak.
I would rather think of it as being around long enough to know what I like and don't like . EQII rntertainerd me 3 months at least, WoW less than a month.
Back to topic though, so Owyn what you are saying is that as long as the cities and countryside remain fairly unpopulated the lag should be minimal?
I will agree that these games tend to be more laggy in crowded area. Ironforge in WoW is a good example. However I also remember Tazoon in Horizons. Because of the way the buildings were rendered that area was the laggiest area of the game and it sure was not because of overcrowding
You could be the only PC in the area and it would still lag. Also on the Horizons note. One of mine, and a lot of others, complaints was running through vast areas with low mob population density in that game. One can only admire the virtual scenery for so long before it becomes a bore fest.
I miss DAoC
You can certainly get client side lag from being in an area with too many polys, too detailed textures, or just poor modeling for the engine (some engines object to specific model flaws or designs, which can stall/bug/mess up rendering). A single character in a big city with a few hundred thousand polys is sure to cause issues for some games!
But in general, that's not a big deal for MMOs. When it is, it's usually either too low a graphics card for the game, too high settings for your card, or poor game level or game engine design on the devs part.
Serverside lag is a more serious issue, because you can't just have the engine reduce poly counts to fix it. You can't fix it by having modelers clean up their models better. You can't fix it by upgrading the client computer. The only ways to reduce serverside lag caused by many players in one area is to upgrade the server specs or improve engine code, and even those hits diminishing returns after a point.
People cluster. The more people on a server, the more clusters you get, and the larger those clusters become. If you scale from 2k players to 10k players, expect the cluster regions on the 2k server that were 50ish in size to become as large as 250ish in size. And all those 10-20 clusters become 50-100 in size. The cluster increase scales pretty evenly, but the server load increases almost geometricly - a 250 player cluster eats up many times as much space as five 50 player clusters, for instance.
That's why it's vital to work at spreading players out through design as you scale up server size.
Owyn
Commander, Defenders of Order
http://www.defendersoforder.com
It's hard to live and easy to die..
I miss DAoC
Absolutely.
I always support the activities of smaller devs who are willing to try to put out a product that's unique or innovative in some manner. 
Owyn
Commander, Defenders of Order
http://www.defendersoforder.com
Uhmm???
Polygons and textures have absolutely nothing to do with server stress. Those things are generated by your system. The only thing that stresses servers is the informatzion that is exchanged between the client (you) and the server. That's things like character level, experience gained, your location on the map and lots of things like that.
You're confusing client lag with netlag.
Graphics have absolutely nothign to do with server lag. You could make a MMORPG with Doom3 liek graphics. That's no problem.
It's hard to live and easy to die..
Uhmm???
Polygons and textures have absolutely nothing to do with server stress. Those things are generated by your system. The only thing that stresses servers is the informatzion that is exchanged between the client (you) and the server. That's things like character level, experience gained, your location on the map and lots of things like that.
You're confusing client lag with netlag.
Graphics have absolutely nothign to do with server lag. You could make a MMORPG with Doom3 liek graphics. That's no problem.
Server databases also handle a few other things also such as the xyz corodinates of you and all the other players in game. Postions of every mob. Damage tables, and a heck of a lot of there things. Not that it matters, technical difficulties mean just that. Technical, not software bugs. The servers have fallen and may never get back up. Ask anyone who betaed Wish about that.
I miss DAoC
well, i beta-tested Wish, and after a few days the lag had been pretty much sorted out, except for in the two big cities.
one server is the way forward, to make a truly memorable game where the players actually have an impact.
[we need a worldwide genocide/ a planetary suicide/ and when the whole damned world is dead/ there''s your f***ing peace]
Not til the tech gets a lot better.
WOW has well over 100k players online at one time during primetime. No idea of the total number - but it's huge. Can you imagine trying to smush them all into one server?
You can't. You'd inevitably end up with clumps too large for any server to handle, and tons of clumps too large for clients to render effectively.
Eve and Wish(?) have shown us that it's possible for smaller games (60k in Eve's case) to have as many as 13,000 players online at one time in one shard. Eve claims to be able to handle 25k at once, which would let them get over 100k subscribers if it's true. I'm not so sure they can handle the clustering at that number, but it could be.
Point is, the Big Games want more than 100k subs, and certainly more than 50k-60k. So until we're able to see 50k-100k players online in one shard at one time (which likely means clients that are able to render several hundred players onscreen at once without severe framerate loss), this won't really be the way forward for most major games. It is self-limiting. A game that launches using a single shard cannot go above 100k or so subscribers, so far, anyway.
For smaller games that focus on niche markets and want to market a very story-based world, however, this is a great way to go.
Owyn
Commander, Defenders of Order
http://www.defendersoforder.com
The lag was bad the first few days because everyone was logging in to check out the free open beta. The lag went away because the players went away, not because they fixed anything. Server technology has a long long way to go before the one world concept will be viable for a mainstream MMORPG.
Although I will agree that players having a impact is the way to go in future games. The sandbox idea ids dead. Horizons had the right idea with players driving back the withered aegis, although their implementation sucked.
I miss DAoC
The lag was bad the first few days because everyone was logging in to check out the free open beta. The lag went away because the players went away, not because they fixed anything. Server technology has a long long way to go before the one world concept will be viable for a mainstream MMORPG.
Although I will agree that players having a impact is the way to go in future games. The sandbox idea ids dead. Horizons had the right idea with players driving back the withered aegis, although their implementation sucked.
I've been in Wish beta too, among other betas. Wish had a different technology than DnL (which uses BigWorld) The problems Wish had are not the same problems that DnL has.
Nemo sine vitio est
Not til the tech gets a lot better.
WOW has well over 100k players online at one time during primetime. No idea of the total number - but it's huge. Can you imagine trying to smush them all into one server?
You can't. You'd inevitably end up with clumps too large for any server to handle, and tons of clumps too large for clients to render effectively.
Eve and Wish(?) have shown us that it's possible for smaller games (60k in Eve's case) to have as many as 13,000 players online at one time in one shard. Eve claims to be able to handle 25k at once, which would let them get over 100k subscribers if it's true. I'm not so sure they can handle the clustering at that number, but it could be.
Point is, the Big Games want more than 100k subs, and certainly more than 50k-60k. So until we're able to see 50k-100k players online in one shard at one time (which likely means clients that are able to render several hundred players onscreen at once without severe framerate loss), this won't really be the way forward for most major games. It is self-limiting. A game that launches using a single shard cannot go above 100k or so subscribers, so far, anyway.
For smaller games that focus on niche markets and want to market a very story-based world, however, this is a great way to go.
Obviously you're not gonna fit all of WoW in that small world. The world there is not made for even 40K+ people.
Dark and Light, on the other hand, is made so you don't have thousands of players cramped in one area trying to level up. It is 40000km squared. The monsters don't spawn in the same spots, depending on weather conditions.
I believe part of their team made their game engine, so it can fit their ideas to create Dark and Light.
http://www.bigworldtech.com/
Don't use WoW an an example. The most populated world is around 30,000 people and the servers truly are crap. This week they moved the most populated worlds to better servers to try and relieve stress but it's only a temporary fix.
The one world idea can work and work now but it depends solely on whether or not the company is willing to spend the the money to do it.