[SOLVED] Viability for a large scale MMO?

[jacek.gadomski]

Hello all, allow me to introduce myself.  My name is Jack and I'm entering the world of game development, after working for a MMO publisher and spending plenty of my own time on contributing ideas/content to other games.

 

I was initially going to use Hero Engine 2 for my project, due to that engine's ease of use and good back-end support.  However, after previewing Unigine I find it impossible to walk away from it.  There are some questions I'd like to get realistic anwswers to. 

 

The Valley Demo actually represents one portion of the world I was in the stages of designing in HE 2, it looks like it came right out of my imagination.  I'm looking to create large open environments like that, with the view stretching for a fair distance.  Is it realistically possible to make a landscape "similar" to that, and have it populated with 100+ NPC's or players?  What am I dealing with here in terms of splitting the area up so it doesn't overload the servers or lag the player's PCs'?

 

The world looks fantastic, I'm just really concerned about performance issues when large amounts of people start showing up and performing actions. 

 

I know there are some MMO's that are using Unigine, but I haven't seen anything like what I'm talking about. 

 

Cheers,

 

Jack

[ulf.schroeter]

UNIGINE provides outstanding graphics quality and with some upcomming SDK extensions unmatched large-world support (with double precision requirements and worlds of several hundred square km).

 

Nevertheless - just by having a 2-min look on the Hero Engine homepage - it looks like it is really optimized for MMO style games and already includes all required AI, server backend etc. This functionallity is not available in UNIGINE yet and would have to be implemented on your own (which might be quite challenging).

[binstream]

UNIGINE is a great technology for a MMORPG client application, while you will need to use other solution for server-side, as Ulf mentioned.

[jacek.gadomski]

Hello, thank you both for your quick responses.  Hero Engine does have that nice back-end like you mentioned, and it is cheaper.  However, the game world I really want to create does not seem possible on HE, with things like realistic force fields, rope mechanics, and all the other bells and whistles that Unigine comes with.  I suppose some of that could be programmed into HE 2 if I had the source code, but I believe it would be as hard as using Unigine and adding another back-end onto it. 

 

On that note, I would like to know what advice anyone here could offer me for a good, solid back-end technology that would balance the server loads properly.  Someone recommended RakNet to me, while another developer suggested HE 2 or even BigWorld.  I feel it's good to get more opinions though if possible.

 

Cheers,

 

Jack

[ulf.schroeter]

RakNet is 'just' a network communication middleware which is/has also been used by UNIGINE for low-level LAN/WAN communication (though they are moving/have moved to their own implementation, I don't know the exact status/roadmap, maybe UNIGINE can also comment on this aspect. Therefore I don't think that this will be any alternative at all.

 

Regarding all the eye-candy UNIGINE can provide (your force fields, rope mechanics, etc) please keep in mind that this works perfectly on a local client machine, but is not designed for network-synchronized operation (which is a totally different challenge due to required excessive state transfer especially for such 'detailed' effects) Just as a small example of these synchronization challenges see this post in the context of multi-channel effect synchronization which is in a way comparable (though even MUCH easier as synchronization only has to take place in a high-speed LAN environment with much lower latency and bandwidth limitations than internet WAN connections, requiring things like local prediction/correction to allow smooth synchronized animation..you can find tons of endless articles on the internet dealing with the challenges of nice WAN network multi-player architectures. 

 

Also have a look on this post for some discussion of the complexity of trying to develop your own MMO solution. Key summary from Alexei here is "you are in the very beginning on the lo-o-ong MMO server development road..."

 

Knowing that he is engaged in MMO backend development (his company builds the RunServer MMO middleware, maybe you should have a look) I would doubt that doing this backend develeopment all alone would be a wise decision...

 

Same applies for the modelling/content side of a large-world MMO environment. you should ask the UNIGINE content team, how many modellers were engaged with the modelling of the Valley demo (or all other demos) and how long it took them to create these environments with such a high quality :) Just by following the announcements/disussions her on the forum I know that it took quiet a long time

[jacek.gadomski]

Thanks again, I will check out RunServer as it looks interesting.  Sooner or later, this level of detail will come to MMO's for sure.  Player computers just need to get with the times.