P2P servers. Open universe.
- Thread starter Greg
- Start date
You know, if Frontier didn't treat commanders as special flowers, but everyone was a solid square/ solid triangle, how much different would it feel? Solo/ Open would ostensibly feel much the same. We would feel like we are part of a greater whole, not a special flower that's hollow. You could still see who was a CMDR, or not. And you could still elect whichever mode worked for you. But we'd not be the glaring exception. Just part of the rule.
Instead, we have a game that scales out massively, but it doesn't scale within massively. It's been labelled an MMO, but Elite really isn't. It's more just endless copies of small game servers running in parallel.
The bones are there for some great possibilities; but again, I just don't believe the developer is motivated to have that 'mega-server' style environment. Just pockets of flowers doing their thing, mostly oblivious to each other (unless one religiously offends the other, then holy god; they make the templar knights look like mother teresa by comparison).
By the way if you want a game whitch is open universe twitch based and is as old as EVE ( and even older ). Look at vendetta online. I played alot of it. It has a small player base with a single universe that can handle hundreds of players in a single sector. This game is runned by a team of 4 people and it had dynamic economy, player capital ships ( those are extremely hard to get ). Its free on mobile but you need to pay a small sub in pc, Its cross platform in a single uni. The game is on greenlight and the devs try to make it look better to attach it to steam. https://www.vendetta-online.com. Graphics are far more below ed but the whole concept is cool.
If we made enough fuss about this, and asked for it as a serious alternative to space legs and possibly atmo landings, it could probably happen. Unfortunately, I believe that we are now stuck with, as Stitch correctly pointed out, a game that has missed its potential by a wide margin.
Never mind, it's still good and as internet lines improve around the world, it should be less of an issue.
Never mind, it's still good and as internet lines improve around the world, it should be less of an issue.
Last edited:
1000 commanders in a single instance is often mentioned with EVE in mind (because it works there, more or less). People tend to forget that the flight model in ED is a totally different ballpark than EVE. It's a pity we couldn't have a quick test, just for fun, with 1000 original ED ships on the EVE workstation. THAT would be a slideshow, I promise you...
People often label P2P as a simple budget decision from FDev. I'm still confident that there's more to it, mainly technical reasons. If anyone could point me to a game with a similar flight model than ED that has 1000 ships in one instance, only then I would start to believe it. Until this day will come I call it all 'wet dreams'...
You are right, and I agree.
Also remember that for things to work with all those players on "same server" they use time dilation. Talk about respecting players time?
EVE was great. Now its old news.
I have prepared video for such posts but for now I will say this - forget urban myths about how superior client/server is over peer to peer. Each of approaches have their limits and trade offs and those at other side - client/server - really aren't those you are thinking about.
With all technical knowledge and despite all peer to peer bugs FD made right choice there.
Client server would lead to more fragmentation than we have now. Localized severs are absolutely necessary for a game like ED, with relatively high relatively high speed movement.
A US and a EU server would not cut it. Living close to the server would give a huge advantage.
P2P with all it's issues, provides the shortest average connection distance and thus probably the lowest average latency. It also evens out the latency between the players in the the instance, making it 'fair'.
Improving the P2P network code to allow for more players in each instance, is probably a far better option than other types of architecture.
Friends lists, Wings, block lists, the upcoming Squadrons and speed power creep are all negatively affecting the 'ideal' matchmaking, so this is no simple task.
ED is it's ow beast. There is no point in comparing it to 'turn based' games like Eve or traditional MMO/FPS games with a bunch of isolated servers. A hybrid solution where a server takes over as instance host when the P2P spagetti gets to complex, would be interesting though.
A US and a EU server would not cut it. Living close to the server would give a huge advantage.
P2P with all it's issues, provides the shortest average connection distance and thus probably the lowest average latency. It also evens out the latency between the players in the the instance, making it 'fair'.
Improving the P2P network code to allow for more players in each instance, is probably a far better option than other types of architecture.
Friends lists, Wings, block lists, the upcoming Squadrons and speed power creep are all negatively affecting the 'ideal' matchmaking, so this is no simple task.
ED is it's ow beast. There is no point in comparing it to 'turn based' games like Eve or traditional MMO/FPS games with a bunch of isolated servers. A hybrid solution where a server takes over as instance host when the P2P spagetti gets to complex, would be interesting though.
Eve doesn't support real time combat. They use a feature called time dilation which slows down time when too many players are at the same spot. Also the reason many players bought ED is because it's nothing like Eve. A central server would only help those who are interested in PvP and Guilds, both features that the game isn't focused on according to David Braben.
A number of issues.
1) Its not full P2P, there are servers. Its a hybrid system.
2) Full C/S would cost more. Possibly necessitating monthly subs or the game going Pay2Win..... not everyone would be happy with that.
3) Ping times. At least with P2P in a 1v1 situation, both are on equal terms in relation to ping. With C/S, the advantage goes to the person with the better ping.
4) Making it like EvE in terms of number of players in battles or whatnot would be a serious technical challenge and require some really impressive infrastructure. Its what Star Citizen is claiming it will be able to do (one day)... its not easy. Games like EvE get away with it because a) its point and click, requiring a lot less network overhead and b) they still have to use time dialiation when things go crazy... not really an option in ED. There are games that are doing closer to what ED is with lots of players, but that is very much that they have been built from the ground up to do it, eg: Planetside 2.
There are some advantages to the P2P system we have though. Lower running costs means either more money for development or less risk of P2W/subscriptions, which is good for the game in general, plus for the longetivity of the game. The higher the running costs, the more dependent it is on maintaining higher player numbers, then you run into the issue of the devs basically having to bow to the vocal majority, turning a unique game into generic FPS #342343 because game X has Y so this game must have Y. Some people would see this as a good thing, personally i like how FD do things differently... if not always to my taste. It gives us something unique. Plus if the majority are crying out for Quake in space, then the poor explorers are going to be left without updates. On the other hand, if the majority are crying out for exploration, then those who don't do exploration will be left in the lurch.
Sure, there are some advantages to a full C/S architecture, but there are also plenty of downsides.
Its probably best to think of ED as a single player game with the possibility of multiplayer with some MMO elements, and a chance for coop or PvP.
Last edited:
Has ESO solved the network problems?
I remember when I played ESO there where horrible lag problems with "zerg trains" being able to induce so extreme lag that everybody else basically had the game paused. On top of that those who know how could create that lag without getting affected (gaming the net code).
In short at the start of ESO the PvP zone was primarily a lag fest - and that with a non-twitch-combat system.
The PvE world was a fake open world - their "mega-server" was a system that created the illusion of a "mega-server" while only allowing a "few" players into the same instance.
(And then the endless discussions about US players playing on EU PvP server and EU players on US server for "night-capturing").
The FD video was fascinating. Your posts on the p2p issues also interesting. Unfortunately, both way over my head on the technical issues, but I did get a sense that FD had to overcome some big hurdles to pull ED together and turn all of the MMO geek stuff into pretty pixels, which makes me less inclined to whinge about the game cos it doesn't have this or that. [big grin]
Last edited:
Honestly I think if complainers actually watched the video and had a think about it they would agree that this game is no simple matter...
.
TL/DR:
.
A switch from P2P to client-server infrastructure is not hard from a coders point of view. It's the accountants who actually flinch at this idea.
.
.
Now the longer version:
.
Switching something from P2P to fixed servers can be done as 30 minutes hackjob. Keep in mind, some P2P Projects at some time analyzed their network traffic and found out that it actually looked not like P2P but like a client-server infrastructure. And there was nothing wrong with their implementation, but it actually was the result of high quality coding. The software just "saw" that some nodes had much higher performance and better network connection, which turned them into central nodes. Thus the analysis of the network made them look like they were actual servers.
.
Now let's translate to ED. Setting up some extremely powerful P2P clients with good network connection could possibly turn them into a kind of "server" organically, but there'd be no guarantee that they'd stay that way. Thus this would not be reliable enough. But unless the whole thing was designed and coded under the influence of drugs and alcohol, there's a function which is responsible to tell each client where to connect to. Change this function to always return the same server instead, and you've effectively switched from P2P to a fixed server. Just don't expect any reasonable performance yet and be prepared for some disconnects and capacity limit problems.
.
To actually get things running properly is more work. It requires to disconnect the current game+P2P combination and create as standalong P2P backbone, which can run without the client. While far from perfect, it would be a good foundation for the server software, which could be run on a server cluster. We know for sure that handing over between different P2P clients at the moment works, the very same handover can also work on a server infrastructure. The big open issue would be load balancing. But based on all we know about ED and how its P2P works, it already has some load balancing included.
.
So all in all in from a software and development point of view, the switch from P2P to a client-server
infrastructure is really rather easy. The switch from client-server to P2P is usually the hard or even impossible one. There's only one elephant in the room, why FD will never do it: the servers and Network connection cost money.
.
I mean, a number of MMOs out there show that it's quite possible to run servers on a B2P model. (Most famous: GW2. ) Considering that server costs scale directly proportional to player numbers while network traffic scales exponentially proportional to players active in one instance, it at first glance seems like the example of GW2 is insanely more expencive to run. (Way less players per instance than GW2. ) But when you also consider the nature of the game, expecially how vulnerable the game is to latency issues, it becomes clear that the smaller instances are not an advantage but merely a necessity. So in the end i would expect EDs costs for running a client-server architcture to land in the same ballpark figure as some of the better MMOs around.
.
Mind you, this whole thing is a list of (more or less educated) guesses. While i have experience with client-server infrastructure and applications, the last time i was involved in the creation and operation of an online game (at that time still text based) was almost 20 years ago. So there might be costs i am not aware of, which would make ED on a client-server infrastructure prohibitively expencive. It might be that FD makes significally less money per customer than GW2, i could not tell. But under the assumption that income per player is similar, FD could definitely afford the switch to a client-server infrastructure.
.
And to also take the video into account posted in this thread: this would actually not be affected by the change at all. The whole video explains how the background simulation works, and that system would work exactly the same, no matter of the infrastructure used for the interaction between players. (Depending on a few internal details, it could even be that the clients would still be able to directly draw info from that system, although it's more likely that the central servers would have to pull and distribute the information. )
.
So all in all, from a technical Point of view the switch would be comparatively simple. It would "only" cost money, which FD obviously doesn't want to spend or can't afford to spend. The important question really is: would a switch to a client-server infrastructure improve the players experience to a degree that it significally affects customer retention. Which boils down to: would the switch make FD more money than the servers would cost them? And apparently FD doesn't think so.
.
Methinks Sylow did not watch the video!!
Having also worked in software development, I must say that I *highly* doubt this level of change would literally take someone 30 minutes. Even if you didn't count all the time it would take to physically set up appropriate server architecture and just focused on making code changes, there is surely enough code behind the game and its networking that even making just a few minor changes would take significantly longer (not to mention doing QA and bug testing afterward). In my experience, going back and "hacking" software into something it was never designed to be usually ends up being trickier than just creating it that way from the start. Not impossible, obviously, but I wouldn't say easy either.
That said, your point is well taken. In principle from a software engineering perspective it may not be all that difficult in comparison with some other ambitious directions (such as atmospheric landings), but it would still be a mistake to trivialize the amount of general *work* (which is not just a matter of coding) it would take to pull this off. In that sense, I stand by my original point that it would be too difficult to justify from FD's current viewpoint.
You're right on here though. Even setting aside all questions of technical viability and difficulty, the basic fact that it would cost real money for a debatable degree of "improvement" (however we define that) means it is unlikely to happen simply for financial reasons. Again, things might have been different had this been the path from the get-go, but I still think the additional costs of maintaining the more robust server infrastructure would have been more expensive overall than just going P2P. Perhaps on the order of other MMOs but its all speculative in the end.
nope.
in case of all those "peer2peer" games that you get named when you ask here "name me one succesfull online multiplayer shooter that is peer2peer", that may be true - simply because those games are (at least up to now) are fake peer2peer games,
using a game engine that only support server2client and was just adopted to have a "dynamic server".
reverting that is of course just a matter of a few minutes.
in case of elite dangerous, we are speaking about a game engine that got multiplayer just added for elite dangerous.
changing that one to server/Client is a bit more fundamental work. You may want to read the "Factorio" DEV blog from more then a year ago, how much work they had switching from peer2peer to server/client with their game.
I have not had issues with ESO or GW2 in the last year. I have a very fast connection though.
Forgive my ignorance on coding a transition from p2p to CS. Given the myriad of potential instances within ED, i can understand how CS could be cost prohibitive.
Would it be possible to have pocket CS instances? For example, a dedicated CS for CGs. When instancing into a CG system, a dedicated server capable of hosting large number of players would be available only for that system. Would this diminish the prohibitive costs for establishing a game-wide CS set up? Would this hybrid architecture be too tough to build?
Would it be possible to have pocket CS instances? For example, a dedicated CS for CGs. When instancing into a CG system, a dedicated server capable of hosting large number of players would be available only for that system. Would this diminish the prohibitive costs for establishing a game-wide CS set up? Would this hybrid architecture be too tough to build?