49 replies to this topic

[ville-v]

starstutter said:

But there's always a catch 22 here. Yeah, you can support that many online, but then it's not an MMO.

Then no MMO except Eve Online is MMO. Unless I am mistaken, no other game has over 2000 players online in the same world in the same time, more like 500-1000.

starstutter said:

Now, I'm no server expert, but if Valve has trouble running 32 players in the same game without severe lag, I don't know how indies think they have a prayer of supporting upwards of 1000 in the same game.

In FPS, one who wins is only about the ping, so it has to be as small as possible. In MMORPG, on the other hand, 100ms is enough for all playing situations, 500ms is enough for most but might seem a little laggy, and even lag of 2000ms would be enough for everything else than fighting. After all, most players will be either walking around, marketing or chatting.

Global chat channels and auction houses are good features lag-wise, because that means there are not certain places, such as pvp arenas or town centra, into which people will be collecting to sosialize. If map terrains are saved locally, only thing to be sent about the surrounding world is information of the nearby actors and their actions.

Many years ago people could not even dream of server hardware which is possible to obtain for $1000 today, yet there were UO, EQ and Tibia.

[alphadog]

rouncer said:

starscrewer-> just cause your a regular ass sniffer, just prooves that you got what you didnt invent.

You sound like a petulant child who has recently been reprimanded, and has a really nasty "I'll show them" streak in him now. Also, you have my vote for DevMaster's "Most Likely To Go Postal".

While starstutter is a little aggressive in style, he usually gives bang-on advice.

Really, every thread you are in, you take people who give advice based on taking a measured approach to learning game development and think they're dream killers, then proceed to bash them. I've seen this in many threads. Why? What is it about having that opinion like that that would make you lash out at starstutter?

Seriously, there should be more game devs. But, the world is full of MMOs that look alike.

Want to see an interesting indie dev who isn't working on an MMO? How about the guy who created Braid? (http://braid-game.com/)

[alphadog]

jakt said:

I was merely contrasting the amount of work involved in creating a single player game compared to a MMO on a given scale, and was interested in why the community believes this contrast to be such a large gap.

You may have missed my own answer in the thread, but I did list a handful of a many reasons why code-related development of an MMORPG, as compared to a single-player RPG (SPRPG?), and without even considering larger level issues like testing and go-live, is much harder.

Security needs, networking complexity, robustness requirements, etc, etc, etc. Some go from non-existent in a SPRPG to highly critical in an MMORPG, like security. Other are just "escalated" in need and complexity, like robustness. (If an SPRPG crashes, only one player is affected. If an MMORPG crashes, so does your company.)

[starstutter]

alphadog said:

Security needs, networking complexity, robustness requirements, etc, etc, etc. Some go from non-existant in a SPRPG to highly critical in an MMORPG, like security.

A good example of this is Stick Arena by xGen Studios. It's not an MMO, but it is a multiplayer flash game powered by an MMO type server (mmocha I think). It used to be a fantastic game when it first came out, but the servers had too many security flaws. Even though I'm sure they did the best they could, the game soon was flooded with so many hackers that it was completley unplayable. It was so bad that it got to where even the hackers couldn't play well because there was so much tampering with the mechanics, people constantly got stuck in walls. If xGen, a company with years and years of networking experience, can't make a security system that can stand up to small time hackers, I would think it would be impossible to create a security system that could stand up to a good hacker determined to bring down the giant (your MMO).

[rouncer]

VILLE V YOUR RIGHT.

4 quad cores linked up will give you tonnes more.

and thats doin it kickass.

[alphadog]

ville-v said:

For 1000 players online, one good decicated server is enough, even though you won't get as good ping as some FPS would require. If you have 100 players online, even some donated server is more than enough.

Could you describe the system (software + hardware + connectivity) you benchmarked to arrive at those numbers? I'd love to actually get a sense of "real-life" numbers. I agree you can do quite a bit with a modern-day COTS server; in fact, I just built a quad-core, 8GB RAM, 1 TB server for about $600. This server would have been drooled over four years ago. But, it's not just the "4 quad cores linked up will give you tonnes more" that is the most important. I'd think available bandwidth is a bigger issue, and that's still expensive these days, relative to COTS servers.

[TheNut]

Well, I suppose if an old P2 233 is still used to host Hotmail's millions of users today, I guess today's system should fly pretty well in comparison ;)

Regarding the OT, I'll just add that when you actually analyze the full set of requirements in a product, any product and not just MMO, you'll realize the fine details make out 80% of the time required to develop it. It's easy to look at something and say "Yeah, I'll build a TCP/IP library that can support up to an infinite number of users" and yes, it's really easy to do. Clustering, adding gateways, backups, synchronization, security, databasing, etc... consume a tremendous amount of time. And if not designed carefully, by the time you're nearing completion the defects will start to roll in. I wouldn't be surprised if WoW devs want to hang themselves with the amount of boring defect work they have to maintain. A lot of would-be game devs really don't look at it from that angle.

Having said that, I think it's fair to consider reducing the amount of M in MMO. There's a certain threshold I believe that is doable, even by just one person. If M < 128 (for example), it's possible, even with very simple and ad-hoc designs, to maintain a usable system. If you negate the use of clustering and multi-network architectures, the product model is simple enough that I'm sure experienced programmers are more than capable of handling. But as the networking and related requirements are reduced, there's a balancing act where the quality in graphics and math are vastly increased. So either way you look at it, you're going to be spending a lot of time somewhere ;)

Oh, and a recent discussion on security. FYI, it's incredibly easy to bring down a server. Finding hole in the protocol is one way (and usually not hard), sending garbage I find has crashed servers numerous times. I've spoken with some buddies that actually executed code on server machines due to faulty message handling. One more reason to work extra hard to prevent that from happening.

[alphadog]

TheNut said:

Well, I suppose if an old P2 233 is still used to host Hotmail's millions of users today, I guess today's system should fly pretty well in comparison

Are you being facetious? I think (hope?) so, and that I missed it. Hotmail does not run on a P2. Back in about 2000, at about the time of the conversion from unix-based systems to W2K, I remember something like ~100 mil users, with something like ~3000 front servers, and a lower number of backend servers, about 1000.

To constrain this issue as it applies to coding, we can see some differences (the "gap") between an SPRPG and an MMORPG. In the latter, you start having to make design considerations which is what I think you are trying to say: do you code up for multiple nodes with a limit of players? Do you go for broke and build a massively scalable system? Just what hardware will be available in a year, or two, or four, and how many more players can you glom onto your 32-core super-parallelized P8?

With an SPRPG, this is simply not as complicated to set in paper...

[ville-v]

alphadog said:

Could you describe the system (software + hardware + connectivity) you benchmarked to arrive at those numbers? I'd love to actually get a sense of "real-life" numbers. I agree you can do quite a bit with a modern-day COTS server; in fact, I just built a quad-core, 8GB RAM, 1 TB server for about $600. This server would have been drooled over four years ago. But, it's not just the "4 quad cores linked up will give you tonnes more" that is the most important. I'd think available bandwidth is a bigger issue, and that's still expensive these days, relative to COTS servers.

Since I do not have a MMO server running, I am basing the fact to the statistics of the Eternal Lands.

]The new server is a Dual P4 Xeon 3Ghz The old server is a Dual P3 1.2 Ghz[/quote] The same, about the new server: [quote]We run all kind of stress tests, and noticed that with 2K pending connections, it went as high as 50% CPU.[/quote] Then some blog posts about the old server (Stats above): [url=http://eternal-lands.blogspot.com/2007/01/yet-another-new-record.html]850 players online, server load stays under 20%.[/url] [quote name= said:

<graphs snipped, follow link>

To be fair, the "before" one was taken in the weekend, where we had about 100 more connections. However, notice how now the usage is at 1/4 of that (with about 570 peak connections). The actual server process took about 6.5% of the CPU with 600 connections.

750 players online, highest peak in bandwidth was only 1.1 Mbps (out).

By that I am estimating server with some cheap nowaday's hardware and 1 Mbps out could happily manage 1000 players.

[gillvane]

At MMORPG Maker, people are still plugging away at making their own MMORPGs. Some people are doing alpha tests, working on models, etc.

There are various Torque projects, and soon RC2 will release source code so people will start adding to that, and probably make some kits similar to the Torque Kit.

If you've got info about making an MMORPG and would like to share it, please drop by. The site has been around for a good while now, (since 2005) and we've managed to put together quite a few folks that are interested in making their own MMO's.