UOX3.org

What is the size of shards that emu's should be targetting? I always thought one was lucky to get 40 or so players online at one time, but perhaps I am wrong. Remember, I am looking at the average, not the few uber shards. It always seems emu's want to design to compete with commerical ventures of thousands of players. I just don't see that being a realistic user base for the majority of shards.

I've never aimed towards UOX3 supporting thousands of users, at least not concurrently online users. However setting our sights high can make for a smoother experience with a small amount of users as well.

As UOX3 stands (being moderately buggy and not very well optimized towards a large user base) I would say your thought of 40 users online at any given time is a rather close guess. I wouldn't shoot for too much higher without alot of work going in to optimizing and streamlining things (particularly our possible overuse of certain packets).

I think it depends a bit on each individual shard. For shards with specific target audiences, like enforced roleplaying servers, a very high number of players is likely to lead to problems. Crowded events/quests, keeping out people who're just there to ruin for others, etc. With an "everything goes"-shard, on the other hand, you get several groups of players, and a higher number of players would make it easier to find likeminded people on the server.

And it also depends on the size of the map a shard uses. An "OSI clone" shard with all maps enabled might seem very empty with only 40 players online average, seeing as they might be spread out over more than 15 towns, numerous dungeons, etc. And with easily accessible recall/gate spells, they'd be hard pressed to even encounter other players randomly.

Using a smaller custom world with fewer towns/dungeons, in-town-housing and possibly restricted fast travel, less players are needed to "populate" the server.

And of course there are completely different ways of doing shards, like having a whole world, but only a small group of players led on a series of quests/missions by a GM, basically having the entire world exist for them alone.

Not everything is about having 5000+ players online at the same time =)

I believe that there should be no fixed limitation on the number of online players – only hardware/speed limitations.

So, for example, if you had a 150Ghz super computer, you could have 80 players in UOX instead of just 40… but most people, on crappy 1.5Ghz computers, will stick to the usual 10-20.

If there is a fixed limit (say, due to IDs or something), it should be higher than the most populated commercial server – but that does not mean it should necessarily sacrifice features or worry about optimizing it so that it runs smoothly at that level.

I agree, and would like to mention there are no fixed limitations that I know of (beyond the logical limits of our storage containers, IE you can only have 4 Billion items and several hundred million characters). UOX3 should not have limits on usage, but it also should decide what the average target audience is.

The difference is while you don't put limitations in place, you do optimize code and packet handling for different situations. For instance, an extra packet here and there doesn't harm much for a shard with 10 people on it. A server hosting thousands at one time would find that extra packet or two as quite a big increase in network traffic.

Some of this discussion is talking about with a current design. Target size influences design. Engineering is all about trade offs, with rarely one totally optimizing for any one aspect (usually long run no one factor has that much influence). But it does, can, and should influcent design (be it trade off complexity, time, maintainabilit, speed, etc). Engineering is rarely just optimization for any factor(lets do the most effeicent design, even if only intended for one user), for one probably, from an engineering point, will sacrifice other things that should have had a higher influence.


So the post wasnt about just getting new hardware, etc. But if an emu was going to be developed, understanding its target audience size helps in PROPERLY setting the weight, when trades need to be made.

On a long-term open-source project like UOX, I believe it balances itself out automatically over time to the audience that uses it most. After all, if a feature is added, and 20% of the users complain of slowness, that feature is then either optimized or removed. At the same time, if there are big features that could be added and would only make it too slow for 2 or 3% of the current audience, then the majority will demand their inception.

It balances itself out. Right now, the average UOX shard is extremely small, so the demand for speed is not there, and so development time is not being focused on it nearly as much as it would be if there was more demand for more players.

Then again, there is also the fact that capability determined audience, too – so someone wanting to make a large shard would not use UOX, and go with competition, and therefore never influence the direction of UOX. I believe, though, that there are enough people on the border-line between influencing the design and abandoning it that it will, over time, balance out anyway.

Same is true for most aspects of development, except where there just is not the developer man-hours available to back the demands – which is how UOX has been for a while…

Of course, I believe that if you know your target audience well enough to gauge popularity, then certain design decisions can be made early-on that have huge ramifications one way or the other down the line, and will also set the initial audience type somewhat.

The official UO was intended for no more than 300,000 sales, and never more than a fraction of that online at any given time. They really underestimated it, and now it has taken years to rebalance everything. So I agree, it is definitely a major consideration.

Sydius wrote:On a long-term open-source project like UOX, I believe it balances itself out automatically over time to the audience that uses it most.

Perhaps it does, that hasn't been my experience, but I don't claim to have experienced it all *grin*.

None the less, one has to have a starting point to "balance" from. What is the intitial objective (again, because a trade off will be made, and to properly weight the trade, one must understand the criteria. There is no, on result for all situations, that I don't subscribe to).

Thus, even with balance, one starts with a target to make some trade decisions. Which brings back to the original question, How big is the AVERAGE shard in terms of online players?

Well, currently there are only 22 shards on UOGateway which is listed as having more than 50 players online on average. Of those, 14 have less than 100. There are some standing out at the top with 761, 686 and 561 average, but for the most part the entire list consists of shards with less than 50 players online average.

Sydius wrote:The official UO was intended for no more than 300,000 sales..

Err. You meant 30,000? Or 300,000 in total from start of product availability to end of product availability?

I'd say a successful shard would have around 50 players on it. However when designing I would try and scale to atleast the 200 mark. Simply because this won't be the norm, but it gives shards room to grow without having to worry about a performance hit once the 51st player logs on(I'm not sure why it would, but this is just an example).

I'm not sure I buy into the whole thousands of players deal with RunUO. I'm not wanting to bash them but the numbers don't seem to add up.

Another thing about the RunUO tests is the insane hardware they are using to be able to just support 7k players online for a brief period of time. A normal shard isn't going to have the resources to get this sort of hardware. So trying to develop a highly scalable server for this goal would likely be a waste.

Sydius mentioned 1.5Ghz and 10-20 players. I'm not sure about UOX but with Sphere you could get 10-20 players on a 500mhz machine connected via dialup without insane amounts of lag.

The other important thing to remember is the amount of items that are going to be on a shard as a result of the players. This is one huge area where online players tests completely fail. They don't account for the fact that an online player base of around 100 can work the item count of a shard towards the 1 million mark without much work at all.

Punt, I said myself that you need a starting point, I believe – something from which to evolve, since the starting point largely dictates the initial audience.

Xuri, I meant 300,000 sales for the complete lifetime of the project. Not 300,000 people online at a single time, not even 300,000 accounts at a single time, but 300,000 sales of the title.

Zorm, my stats for UOX were exaggerated guesses – me trying to be funny and all. Ha. Ha.