GOD bless Jon and Client/server architecture vs ugly p2p

I can still remember all these RTS alpha i've played implementing p2p model.
It was a nightmare. I was spending much more time trying to connect to a game than playing itself.

This Alpha proves without any ambiguity that p2p architecture for RTS is just old school and should never be seen anymore in any RTS game.

* I've never had a single issue to connect.
* Desynch are old story
* NIL are old story
* slower player making all other players slowing down is old story

If you're not in alpha, then you do not have access to bug tracking, but it's obvious : 99% of reported bug are not about network/connection issue. Remember all these RTS forums were after game release there were massive spamming about players not able to join. Today Uber can concentrate on the game itself instead of dealing with any possible connection issue.

Frankly, i've never seen a so much solid RTS implementation, especially considering we're just at the beginning of the ALPHA.

PA is just about : Join and PLAY !!!!!!

Kudos to Jon and uber team for this revolutionary stuff.

There were lots of promises in Kickstarter. So far i can't see where Uber is failing in delivering what they have promised, and this is really impressive knowing a small team is behind the scene.

I daren't imagine how much time this single team is working hard to deliver this stuff so far.

I only might say the same about flexibility of engine: it's reload everything on the fly: shaders, units configs, sounds, may be even models (I didn't tried it yet). UI is awesome tech which give unlimited possibilities any have no limits at all. Even WebGL working here.

I already think that when server released i will try to make RPG with skill tree/leveling/quests on PA engine, that should be awesome to have RPG with robots.

Client/server architecture only worked with decent development talent and access to startup funds for server hardware/renting (for official servers, of course, player/local servers will be coming which reduces the load considerably).

P2P is dated, but it has it's uses. Of course, there's a difference between the classic P2P model whereby all peers sync data between them, and the emergent P2P model that actually uses a modicum of client-server logic (whereby one machine is the host/server).

I am amazingly impressed at PA so far, but one promising game does not equate to a case study for the industry. If it could, that'd be amazing though

Yeah, it is pretty nice. This is a major advantage over most modern RTS, COH2 for example is P2P and that has been a major complaint since closed beta.

I want this...

Ohh yes Coh2s p2p system is the only reason I wont buy it any time soon.
I love playing 4v4s, but at least in the open beta its almost impossible to complete a 4v4.

On the other hand I never had any network problems in PA so far even though its early alpha.

Playing Company of Heroes 2 also and the connection issues are very annoying and I doubt they will go away any time soon, original Company of Heroes had these problems for a long time.

Battle for Middle Earth also had ridiculous connection issues

yeah that's what I mean, they're problems with the P2P system which is a real shame because it's an otherwise very promising game.

Lets not circle jerk here. I'm not very knowledgeable about this stuff, but the current asynch model used must have some drawbacks. Otherwise everyone would be using it. Unless the tech to fully utilize it has only recently been developed, then I understand.

As I understand it the main idea behind the synchronized model was to make matches more fair by not giving one guy with a faster PC or internet connection an advantage. Back in the days of dial up and sub 1ghz single core CPUs this was a real issue. In the SupCom era it was probably used out of habit.

But now that broad band is so common and even low end PCs are so ridiculously fast its just not much of a problem any more.

The async client/server model has been around since the mainframe era in the 70's and 80's. People just didn't use it for mainstream games until recently. And even then most games that have used it are FPSs or MMOs. Partly because a home PC just doesn't have the horsepower or bandwidth to host an MMO or a very large FPS match and partly because people tend to try and cheat in MMOs and FPSs and the async client/server model makes cheating much harder by virtue of how it works.

You can make it hard to cheat in the synchronized model as well but it requires more work that is not needed with the async client/server model.

Well, maybe you want to inform yourself before you pass judgement on how important this feature is then? The advantages over p2p are immense!



If only one PC in a p2p network structure lags everybody else is screwed too. And every PC needs to able to connect to all other PCs for it to work properly. The chance for that to happen in a 10 vs. 10 match is basically zero.

In return the server is a single point of failure that a p2p network does not have.
But yeah given the circumstances of a game like PA client/server is probably more stable.

This is an excellent point that I failed to bring up. Mesh networks do not scale. At all. Client/server networks on the other hand will scale to the size of, well, the internet.

That's only a P2P advantage if you share non-time-critical data in small packets with a large userbase (torrent network). In your regular RTS P2P network every PC can become a point of failure unless it outright disconnects.

In PA you could just switch regions. If the US server goes down you can still play in EU or Asia. The ping difference would be negligible for RTS. And if all of Uber's servers were to go down at the same time for some reason you could still rent your own server, set up a VPN, play in a LAN or play locally against AI.

This is brand new tech, so now you understand. Chronocam? Who else has this stuff? Nobody else is revolutionizing RTS tech.

This. In fact, nobody is doing anything with RTS tech period, let's be real.

Uber, you guys rock, all of you. Keep it up.

Thanks!

After seeing how well this works, I wonder why p2p is still used????

As far as I know, the tech to do this is very old and has been proven effective time and time again in several MMO games and some FPS games. (Eve online and World of WarCraft come to mind.)

But, I think this is the first time it has been applied to an RTS game and all I can think is that wow, this seems to work so well.

Thanks devs! Keep up the good work.

Hey man, chill out. I specifically said that I'm not very knowledgeable to dilute the strength of my statement. I didn't even make a strong statement in the first place. I understand the basic structure, I just don't know the nuances.

One of the drawbacks I could think of is that if the server goes down, the entire game is gone, but this is also currently present, it's just not as explicit.

My reasoning was that Jon Mavor is not the only rational RTS game developer out there, and if this is one of the few RTSs that uses the asynch model, there must be a good reason for that, hence my previous post.

I was speaking about the asynchronous model specifically. Chroncam is cool, and I don't disagree that Uber is pioneering new ground in RTS. I'm actually curious to see how you guys manage to make this playable and interesting.
I really appreciate what you do and TA was in integral part of my childhood.

The specifics are unknown but from what neutrino has written in the best they do pretty fancy stuff that is really QUITE different from what RTS have done before tech-wise. So they had to
develop new technology to do this, while other RTS just do what they have always done.
I'd really like to eat up an extended long explanation of the way PA does stuff. Please Mavor, write one at some point