Login

russian armor

P2P is already starting to piss me off.

raw
26 Jun 2013, 09:32 AM
#21
avatar of raw

Posts: 644



You mean Warcraft II Battle.net edition? The Battlenet you see there, compared to the battlenet you see now is VERY different.



And at least for me it worked VERY much better than relics P2P system, even back then. :)
26 Jun 2013, 09:50 AM
#22
avatar of cr4wler

Posts: 1164

jump backJump back to quoted post25 Jun 2013, 21:46 PMraw


enlighten us.


most (or at least a lot) of the problems that people describe here would not be going away simply by switching to a server based architecture. on top of that, people don't realise that with the switch to steam, we already have a semi server based system in place.

ping/lag issues? best approach would be a regional setup like sc2 or lol or similar systems. even then you will still have these problems unless you implement it the LoL way (as in, no response from client? who cares? game goes on), which would mean that if your net spazzes out for a second or two (or someone using your connection starts downloading sth or whatever), you might already have been "kicked" from the game (with no possibility of reconnecting, but that could be implemented. but even if it does get implemented, how do you make it work? you have to play against the AI until the VPs run out? His units just stand around? doesn't really work in coh).

drophacking? in a perfect, completely server based system, you should be able to detect who is trying to drophack. in the current state, it is pretty much impossible to tell who the perpetrator is.

maphacking? server based system does nothing to prevent that. your client needs all the information anyway, so the only hurdle to pass is to make that information accessible. a well written hack will be able to fool any kind of cheat detection/prevention.


while i am inclined to agree that the current system is not flawlessly implemented, it is a reasonable compromise (as in getting the most use out of putting in the least possible effort).

also, i don't know if you played on b.net back in the days, but i remember it as being laggy as hell too sometimes, especially if you got a guy that thought it was cool to play in a different region.
raw
26 Jun 2013, 10:24 AM
#23
avatar of raw

Posts: 644

jump backJump back to quoted post26 Jun 2013, 09:50 AMcr4wler

ping/lag issues? ...

drophacking? ...

maphacking? ...



No one here claimed that this stuff doesn't happen in a S-C environment, nor that S-C is the holy grail of networking. This is mostly a QoL issue. As others have already pointed out, relics networking is "meh", to put it carefully.

also, i don't know if you played on b.net back in the days


Yes, and I personally never had any problem with battle.net and that was back in the day when I still dialed in with a modem. Maybe I was excessively lucky or the people responsible at blizzard are just damn good at their job. Fact remains that I have trouble staying connected in relic games despite a uber 1337 h4x0r connection.

Since this hopefully won't be relics last game and they're the second worst at this network stuff right after Paradox Development Studio (who are literally stuck in 1989) I hope someone at relics snaps their fingers and hires a couple of guys to deal with it or something.
raw
27 Jun 2013, 12:43 PM
#24
avatar of raw

Posts: 644

For completeness sake I sat down and counted the games that finish prematurely due to laggers (1v1 and 2v2, not counting 3v3 and 4v4 because it's almost guaranteed a lagdrop). A whopping 40% of my games are decided prematurely, because someone lags out during the course of the game.

This is just awful.
Seb
27 Jun 2013, 12:49 PM
#25
avatar of Seb
Admin Black Badge

Posts: 3709 | Subs: 2

You should look into opening the ports used by the game. It's probably the same as coh1.
27 Jun 2013, 12:56 PM
#26
avatar of cr4wler

Posts: 1164

jump backJump back to quoted post26 Jun 2013, 10:24 AMraw


No one here claimed that this stuff doesn't happen in a S-C environment, nor that S-C is the holy grail of networking. This is mostly a QoL issue. As others have already pointed out, relics networking is "meh", to put it carefully.


i don't understand what exactly you mean by "relics networking".
pretty much everything in the game is done via steam.

i onced dropped in a match because the steam server i was connected to restarted. i got connected to another steam server, but the game ended like a drophack used to end the game (this is vCoH steam version). i'm not quite sure how much of this is relics fault and how much of it is valves, but again, steam provides the network, only thing relic does is call the steamworks API do get stuff done.


Yes, and I personally never had any problem with battle.net and that was back in the day when I still dialed in with a modem. Maybe I was excessively lucky or the people responsible at blizzard are just damn good at their job. Fact remains that I have trouble staying connected in relic games despite a uber 1337 h4x0r connection.

Since this hopefully won't be relics last game and they're the second worst at this network stuff right after Paradox Development Studio (who are literally stuck in 1989) I hope someone at relics snaps their fingers and hires a couple of guys to deal with it or something.


meh, my bnet experiences were mixed... most of the times it was fine, but i experienced quite a few drops and lagfests too (was only playing broodwar though). overall, my experiences in vCoH and CoH2 so far are not that bad (excluding games with guys that literally had an average of below 0.2fps... how bad does your machine have to be to deliver those kinds of numbers? i mean, thats 1 frame every 5 seconds... and i have encountered more than 1 guy with those kinds of numbers). lags and drops do happen though, and would even with a client/server architecture (and i don't think the percentage would drop much).

27 Jun 2013, 13:07 PM
#27
avatar of NorfolkNClue

Posts: 391

jump backJump back to quoted post27 Jun 2013, 12:49 PMSeb
You should look into opening the ports used by the game. It's probably the same as coh1.


I'd argue that most modern routers incorporate SPI, so in theory, port forwarding should not be necessary. For older routers, perhaps still though.
raw
27 Jun 2013, 13:07 PM
#28
avatar of raw

Posts: 644

jump backJump back to quoted post27 Jun 2013, 12:49 PMSeb
You should look into opening the ports used by the game. It's probably the same as coh1.


You should look into actually reading my post. I was counting any drops occuring, not just my own.
raw
27 Jun 2013, 13:08 PM
#29
avatar of raw

Posts: 644

jump backJump back to quoted post27 Jun 2013, 12:56 PMcr4wler


i don't understand what exactly you mean by "relics networking".
pretty much everything in the game is done via steam.


No.
Seb
27 Jun 2013, 13:12 PM
#30
avatar of Seb
Admin Black Badge

Posts: 3709 | Subs: 2

jump backJump back to quoted post27 Jun 2013, 13:07 PMraw
You should look into actually reading my post. I was counting any drops occuring, not just my own.

Then you don't understand P2P, because there is no such thing as your own drop or other people's drop in P2P model.

The only common factor to all of those games is you. And I'm offering advise to help you out.
raw
27 Jun 2013, 13:59 PM
#31
avatar of raw

Posts: 644

jump backJump back to quoted post27 Jun 2013, 13:12 PMSeb

Then you don't understand P2P, because there is no such thing as your own drop or other people's drop in P2P model.

The only common factor to all of those games is you. And I'm offering advise to help you out.


It is news to me that my connection makes other people lag in P2P.
27 Jun 2013, 14:05 PM
#32
avatar of NorfolkNClue

Posts: 391

I think that the main problem is that games have to use UDP to talk to each other. UDP is an unreliable protocol, but for reasons of speed, especially in real-time games like CoH, it is used pretty much exclusively.

So, when a UDP packet is dropped (for any reason along the route), the sending machine has no way of knowing that it has been, and will continue merrily firing off packets to their destination. If there's enough packets dropped, then the receiver will think that the the sender has died. Compounding this if the packet size is larger than normal, then the packet will get broken into more fragments than normal (every packet is broken up so that they sit within the network's configured MTU size, larger packet means more fragments). If any *one* of those fragments get lost, then the whole packet is dropped.

It's actually surprising that any UDP games work at all, ever really.

What I'm trying to say is that the way CoH works is an established method, but is subject to the vagueries of the internet and its routing. If your current route is traversing a poorly configured network, then the chances of you dropping packets increases. The reasons are countless.

As for how the client handles this, that's a different matter, but at some point if it suddenly can't communicate with another machine, you can't expect it to sit there for ever waiting for the traffic that will never come because of some problem along the way that no-one [in this context, no-one is either player or relic. Clearly someone has control over it somehere :P] has any real control over.

I imagine the client's behaviour could be improved, but so could anything. After watching the development of this game over the last few months, I actually give the devs some credit in that if it's something that they can do something about, they will.
27 Jun 2013, 15:01 PM
#33
avatar of cr4wler

Posts: 1164

jump backJump back to quoted post27 Jun 2013, 13:59 PMraw


It is news to me that my connection makes other people lag in P2P.


which leads me back to my initial post.

there is no "your" and "the other persons" connection in P2P. there is only "your shared" connection.
27 Jun 2013, 15:49 PM
#34
avatar of CombatMuffin

Posts: 642

I think that the main problem is that games have to use UDP to talk to each other. UDP is an unreliable protocol, but for reasons of speed, especially in real-time games like CoH, it is used pretty much exclusively.

So, when a UDP packet is dropped (for any reason along the route), the sending machine has no way of knowing that it has been, and will continue merrily firing off packets to their destination. If there's enough packets dropped, then the receiver will think that the the sender has died. Compounding this if the packet size is larger than normal, then the packet will get broken into more fragments than normal (every packet is broken up so that they sit within the network's configured MTU size, larger packet means more fragments). If any *one* of those fragments get lost, then the whole packet is dropped.

It's actually surprising that any UDP games work at all, ever really.

What I'm trying to say is that the way CoH works is an established method, but is subject to the vagueries of the internet and its routing. If your current route is traversing a poorly configured network, then the chances of you dropping packets increases. The reasons are countless.

As for how the client handles this, that's a different matter, but at some point if it suddenly can't communicate with another machine, you can't expect it to sit there for ever waiting for the traffic that will never come because of some problem along the way that no-one [in this context, no-one is either player or relic. Clearly someone has control over it somehere :P] has any real control over.

I imagine the client's behaviour could be improved, but so could anything. After watching the development of this game over the last few months, I actually give the devs some credit in that if it's something that they can do something about, they will.


That's interesting. I'd assume by that then, that UDP is a safer method (since you technically can't receive incomplete packets, otherwise it drops) but it runs the risk of being laggy (or unstable) in a game environment.
27 Jun 2013, 17:07 PM
#35
avatar of nin3

Posts: 59

I think that the main problem is that games have to use UDP to talk to each other. UDP is an unreliable protocol, but for reasons of speed, especially in real-time games like CoH, it is used pretty much exclusively.

So, when a UDP packet is dropped (for any reason along the route), the sending machine has no way of knowing that it has been, and will continue merrily firing off packets to their destination. If there's enough packets dropped, then the receiver will think that the the sender has died. Compounding this if the packet size is larger than normal, then the packet will get broken into more fragments than normal (every packet is broken up so that they sit within the network's configured MTU size, larger packet means more fragments). If any *one* of those fragments get lost, then the whole packet is dropped.

It's actually surprising that any UDP games work at all, ever really.

What I'm trying to say is that the way CoH works is an established method, but is subject to the vagueries of the internet and its routing. If your current route is traversing a poorly configured network, then the chances of you dropping packets increases. The reasons are countless.

As for how the client handles this, that's a different matter, but at some point if it suddenly can't communicate with another machine, you can't expect it to sit there for ever waiting for the traffic that will never come because of some problem along the way that no-one [in this context, no-one is either player or relic. Clearly someone has control over it somehere :P] has any real control over.

I imagine the client's behaviour could be improved, but so could anything. After watching the development of this game over the last few months, I actually give the devs some credit in that if it's something that they can do something about, they will.



probably you are right.
could you explain what happened with coh1? Before relic moved to this new server, I had no problem with lagg, dropped games, etc.
Since the move to this wonderful server, its a nightmare, for everybody, drops, laggs.
Since its still p2p, makes no sense to me, why is is happening.

sry for the highjack
27 Jun 2013, 17:21 PM
#36
avatar of crazyguy

Posts: 331

jump backJump back to quoted post27 Jun 2013, 13:59 PMraw


It is news to me that my connection makes other people lag in P2P.


As has been stated before, it is a shared connection. A drop rate of 40% is abnormally high, so it is very likely your connection has something to do with it.
27 Jun 2013, 18:57 PM
#37
avatar of CombatMuffin

Posts: 642

I am talking out of my ass here, but even though it is a P2P connection, steam may have some sort of validation integrated in the beginning and end of the matches. That may be a reason for the usual "xxx is lagging" at the beginning of every game (the game is handshaking Steam before doing its thing).

Again, pure speculation, but there has to be handshake somewhere, so as for Steam to know how the game went, or if a game even begun.

Regardless: I find it pointless to debate P2P. It is more likely that they'll introduce interactive replays into the game than change the entire networking code for it.
27 Jun 2013, 18:58 PM
#38
avatar of cr4wler

Posts: 1164



That's interesting. I'd assume by that then, that UDP is a safer method (since you technically can't receive incomplete packets, otherwise it drops) but it runs the risk of being laggy (or unstable) in a game environment.


quite the contrary. both is UDP considered "unsafe" since it's a stateless connection, and there is no such thing as an "incomplete" packet. an incomplete packet is a corrupt packet, and a corrupt packet can occur either way (both in stateless and in stateful communication).

little explanation: stateful communication means that you first "establish connection". the so called handshake is pretty much that, a handshake between the two transmission partners, "getting to know each other" and the intention of transmitting data. in stateful communication, each sent packet will be numbered and each received packet will get acknowledged (though they can be acknowledged en bloc). this means the client can tell that a packet if missing if he got packets number 1-14 and 16-30. he can then specifically ask the partner to re-send packet 15. In UDP, there is no handshake, and neither is there acknowledgement of received packets. In UDP you simply send messages to your intended target without regard for loss (actually, it is expected that there probably will be loss at some point). On top of that, since there was no handshake etc. the two transmission partner never tried to find out how fast they might be able to interact with one another and UDP has no inherent congestion control. If your partner sends packets faster than you can receive them, there might be packet loss.
For all these reasons, UDP is mainly used in real-time applications, where outdated/lost packets are more or less irrelevant (VoIP, Streaming, Gaming).
raw
27 Jun 2013, 19:06 PM
#39
avatar of raw

Posts: 644



As has been stated before, it is a shared connection. A drop rate of 40% is abnormally high, so it is very likely your connection has something to do with it.


It hasn't.
raw
27 Jun 2013, 19:07 PM
#40
avatar of raw

Posts: 644


Regardless: I find it pointless to debate P2P. It is more likely that they'll introduce interactive replays into the game than change the entire networking code for it.


I think this discussion is useful in so far that it may motivate Relic to "change their ways" for their next game. Don't stop believin'! :)
0 user is browsing this thread:

Ladders Top 10

  • #
    Steam Alias
    W
    L
    %
    Streak
Data provided by Relic Relic Entertainment

Replay highlight

VS
  • U.S. Forces flag cblanco ★
  • The British Forces flag 보드카 중대
  • Oberkommando West flag VonManteuffel
  • Ostheer flag Heartless Jäger
uploaded by XXxxHeartlessxxXX

Board Info

966 users are online: 966 guests
1 post in the last 24h
9 posts in the last week
27 posts in the last month
Registered members: 50005
Welcome our newest member, swimmingpoolsofflori
Most online: 2043 users on 29 Oct 2023, 01:04 AM