Forums » General Pantheon Discussion

Player To Zone Ratios: An excercise

    • 337 posts
    September 14, 2021 11:20 AM PDT

    So, Vjek posted a snarky and interesting post earlier that got me thinking about this topic.  Currently there are nine zones and three starting cities on Kingsreach. Extrapolate three continents.

    3 x 9 zones = 27 zones.

    9 starting cities for the nine races.

    I believe I heard Joppa mention once in an Amberfaet stream that he was confident that the content could support 100 players grouped up.  And let's say there are some in there just gather resources or snoop around.  Let's call them 20.  If there actually were 120 players at a given time in each zone, plus an additional 40 in zones with starting cities with an average playtime of 2 hours, how many people could the world sans networking hold?

    ((27*120) + (40*9)) * 12 = 43,200 active players per day.

     

    How does that marry up with the new networking stack plus HDRP and terrain streaming?

    • 1921 posts
    September 14, 2021 11:45 AM PDT

    IMO:

    It's more about prime-time concurrency, rather than per day.  Afaik, they've never posted server population design goals, so it's all speculation.

    100 players grouped up per zone is .. rather low, with only 2 zones per tier per continent, if you expect to have the EQ-esque "reputation matters" server pop cap of ~3000.
    If you're considering only the combat loop, then TTK & respawn rate has a part to play as well.

    • 145 posts
    September 14, 2021 11:58 AM PDT
    They are starting with 12 zones I think so that is 1200 players right? Also I'm sure the team has its own analytics but fun to think about
    • 337 posts
    September 14, 2021 12:05 PM PDT

    Now I am not talking about Alpha, I'm talking launch so that should be three complete continents.  There's nine zones just on Kingsreach if I'm counting correctly.

     

    I'm not smart enough about networking to know any of this, but it seems like 3k per server of EQ was more a limitation of design of networking combined with world size.  Now that tiles are managed more efficiently even within zones combined with more efficient packeting, you'd think that the 3k per server deal would be history.  But maybe I'm speaking out of ignorance.

    • 1999 posts
    September 14, 2021 12:17 PM PDT

    Is one of these 10 places not a zone?

    South Saol
    Vae Woods
    Faerthale
    The Roans
    Avendyr's Pass
    Thronefast
    Veil of Azeris
    Eastern Plains
    Wild's End
    Silent Plains

     

    I'm not stalking you Benonai, honest :)


    This post was edited by Jothany at September 14, 2021 12:20 PM PDT
    • 793 posts
    September 14, 2021 12:32 PM PDT

    If VR truly wants us to create community, then server populations can't get too large. 

     

    Someone smarter than I could probably figure out that tipping point from "enough to create a world that is alive with people" and a "world that we are just blobs of pixels to each other, outside your close circle of friends"

     

    • 134 posts
    September 14, 2021 12:44 PM PDT

    100 people to a zone is a massive amount, not small at all. But Wild's End is 20 minutes across with no obstructions. That's about twice as large as most of the Kunark zones I believe (load up EQ and run across Dreadlands sometime. I think it takes a little more than 10 minutes without SoW to go from the Luclin Spires to Firiona Vie. Most of the other zones are a little smaller.)

    On top of that Wild's End, and probably many of the other outside zones, will actually be the equivalent of several zones in one. For instance, in Wild's End we'll have the Giant Ruins, Well Pond, Sorhiryth, and Mad Moors (I'm forgetting something, I know I am.) 100 people bouncing around all of those areas and you probably won't run into each other too much (post launch day.) 100 people in a zone like Amberfaet or Halnir's Cave would be a different story. Unless those zones are Dreadlands sized with a high mob density, then 100 people in a dungeon will be super crowded.

    • 1921 posts
    September 14, 2021 8:09 PM PDT

    Fulton said: If VR truly wants us to create community, then server populations can't get too large. 

    Agreed

    Benonai said:

    ...

    I'm not smart enough about networking to know any of this, but it seems like 3k per server of EQ was more a limitation of design of networking combined with world size.  Now that tiles are managed more efficiently even within zones combined with more efficient packeting, you'd think that the 3k per server deal would be history.  But maybe I'm speaking out of ignorance.

    IMO:

    It comes down to where you draw the line in your design and implementation, on the server side.  Do you run a unique process per zone, with every mob in it's own thread?
    A* pathfincing on ANY kind of node based or world geometry navmesh is... horrendously unpredictable when you start to get into the hundreds of NPCs (which can all (or none) be active at any given moment).
    Do you run all the zones for one server in one container, one operating system, or other/similar/different virtualized structure? Not always the best choice..
    Do you run all the proceses for one server in one operating system instance?  It's not necessary, but it is convenient if you don't design things to be portable and intended for use at scale.
    DPC Latency and Context Switching become problematic when you put more than 200 active processes all demanding significant time slices from any monolithic kernel at 2-4ghz.  Problematic as in, human noticeable delay of all I/O.  The illusion of multiprocessing starts to break down quite handily. :)

    Making a single kernel handle high load/low latency disk, cpu, memory, and network when your avg target latency for >1M IOPS is sub 1ms, and concurrently, service the interrupts generated by ~saturated multi-gigabit is .. asking a lot.
    Ideally, nothing would run at or near capacity, to keep latency as low as possible.
    Putting your login server, zones, and database all in one operating system instance?  Poor design followed by poor implementation.  You gotta separate all of those.

    Essentially, the only real requirement is to reliably hand off the client connection from one zone to another.  If that's abstracted at the Layer 3, that is, it inherently permits a different IP:Port combination to connect to, then 3000 doesn't have to be a limit at all, because you could have one zone per container, instance, VM, operating system, whatever, if desired/required.  I mean, I personally think that would be a bit overkill, but it would permit tremendous scaling potential. 

    Part of the above is also dependant on the persistence layer.  If your persistence layer is similarlly abstracted, load balanced, and scalable, then again, the only concern for server population, in practice, is server->persistence latency and your network infrastructure supporting whatever packet-per-second rate on whatever interfaces permit the servers to get confirmed writes from the databases in a timely fashion.  Put another way, you would not want all your zone servers in Australia when all your Databases are in Ireland. :)  Regardless of where your clients are connecting from.

    • 337 posts
    September 14, 2021 8:33 PM PDT
    I followed some of that ... Again, ignorant here. But I thought that the special sauce that Kyle was cooking was to have a library on the networking side so that would free up server-side demand and greatly increase the amount of players that could be instanced? They obviously still need to break up the zones but this allows them to make very large zones to hood more players. Otherwise there's no reason to have large zones. As much as too many players on one server can be detrimental, too few players per so acre of zone is also detrimental. That's why I figured they have an idea of how many people would fit comfortably with x amount of space and content then we could extrapolate the size per server they were thinking.
    • 1921 posts
    September 14, 2021 10:25 PM PDT

    IMO:

    If they know, or have an idea or target for that, Benonai, they haven't shared it in over seven and a half years. :)

    The problem of concurrency per zone, if you have one zone per kernel, is a non-issue, because you have the entire operating system available to service and maintain the client connections.
    Do some google searches regarding udp and/or tcp concurrency in any modern kernel and you quickly find out it's >10k connections, handled natively by the OS.
    You're not going to have >10k players in a single zone, so with the proper scale, design, and implementation, it's a solved problem.
    That's why the design is so important.  You don't need anything special or new if you have enough abstraction in your design. 
    I mean, obviously, it was possible in 1999 and things have only become faster, better, and higher capacity in the intervening time (on the server side of things).

    Honestly, I have no idea why they need anything 'special', other than Unity deadlocking the main rendering thread on Network I/O by default. (in the past)
    But they knew that back in 2014, so.. eyes wide open, they stuck with it, even knowing that.. /shrug

    • 2419 posts
    September 15, 2021 7:17 AM PDT

    Byproducts said:

    100 people to a zone is a massive amount, not small at all. But Wild's End is 20 minutes across with no obstructions. That's about twice as large as most of the Kunark zones I believe (load up EQ and run across Dreadlands sometime. I think it takes a little more than 10 minutes without SoW to go from the Luclin Spires to Firiona Vie. Most of the other zones are a little smaller.)

    On top of that Wild's End, and probably many of the other outside zones, will actually be the equivalent of several zones in one. For instance, in Wild's End we'll have the Giant Ruins, Well Pond, Sorhiryth, and Mad Moors (I'm forgetting something, I know I am.) 100 people bouncing around all of those areas and you probably won't run into each other too much (post launch day.) 100 people in a zone like Amberfaet or Halnir's Cave would be a different story. Unless those zones are Dreadlands sized with a high mob density, then 100 people in a dungeon will be super crowded.

    It really isn't about the size of the zone but the amount of available content within it.  In that latest stream Joppa said that their zones can have a height above the center point of 3,000m and a depth of 3,000m.  That's great and all, but if that amount of volume is empty it's basically worthless.  That wellpool concept art looks great, but if it's mostly devoid of anything it too is mostly useless space.  The Karana's in EQ1 where utter shite because, while very large, were also mostly empty space.

    I would tell VR that however much content they think they need in a zone, increase that by 50% and however many people you think that content can support simultaneously, cut that by 50%.  Having more content available than what the population can actively engage is better than the opposite. EQ1's problem was that it kept funneling more and more people into fewer and fewer zones as you leveled up.  Instead of pyramid shape, make it a column.

    If it were up to me, I would ensure that as you go up in levels there is actually more content, percentage wise, across the entire world so as you level your choices actually increase significantly.

    • 3852 posts
    September 15, 2021 8:15 AM PDT

    If ignorence is bliss I must say I am extremely happy posting in this thread. Referring only to my own ignorence.

    I do not know VR's thoughts on how many players they want to have on one server. As noted above too small makes it feel dead and makes it hard to get groups - which is deadly in this particular game. Too large makes community and reputation matter a lot less than many of us want. 

    To make the decision far harder - they need to factor in that a lot more people are likely to create characters on a server at launch than to still be playing in even three months. The "steady state" population of new servers in MMOs tends to be very significantly smaller than the "Woot new server I want to try it" population at launch or even a week later. Thus they may - or may not - need to plan for a server able to hold a lot more people than they really want for the first month or two. While still providing good server stability and low lag. And without excessive competion for limited resources including mobs. Since they *really* want to give new players a good first impression and not have a game that "sucks" for three months until population thins out. 

    I do not know how many concurrent players a server will be able to accomodate without instability, serious lag, and major bottlenecks in access to mobs and resources.

    I do not know the time zone distribution of players. A server with 1/3 on European times, 1/3 on North and South American time and 1/3 on Oceanic time will be spread out over 24 hours and can have more players with less crowding at any particular time. 

    I do not know how mobs, resources and things to do will be spread out over a zone. If 100 players are killing giant rats outside of a city we have a problem. Especially if VR chooses to have pure competition with no form of shared credit for a rat multiple players/groups attack. If a large zone has 50 camp areas and 100 players scattered around the entire zone we do not have a problem. As long as the server can support it.

     


    This post was edited by dorotea at September 15, 2021 8:16 AM PDT
    • 134 posts
    September 15, 2021 8:27 AM PDT

    Vandraad said:

    It really isn't about the size of the zone but the amount of available content within it.  In that latest stream Joppa said that their zones can have a height above the center point of 3,000m and a depth of 3,000m.  That's great and all, but if that amount of volume is empty it's basically worthless.  That wellpool concept art looks great, but if it's mostly devoid of anything it too is mostly useless space.  The Karana's in EQ1 where utter shite because, while very large, were also mostly empty space.

    I would tell VR that however much content they think they need in a zone, increase that by 50% and however many people you think that content can support simultaneously, cut that by 50%.  Having more content available than what the population can actively engage is better than the opposite. EQ1's problem was that it kept funneling more and more people into fewer and fewer zones as you leveled up.  Instead of pyramid shape, make it a column.

    If it were up to me, I would ensure that as you go up in levels there is actually more content, percentage wise, across the entire world so as you level your choices actually increase significantly.

    Wholly agree with this. It was something I was thinking about but didn't get across in my writing. Big is important for exploration, but dense is important for population support. That's what I was thinking of when I said that 100 players is a lot. Even the biggest, most NPC rich zones in EQ could rarely support more than 10 groups worth of players and that was usually stretching it quite a bit. You'd have a lot of friction between groups even if there was a lot of space between them, or a ton of downtime for everyone.

    However as you said, often there was no choice where to go. Everyone remembers Guk or Old Sebilis because, well, where else could you go? Even by the time we hit Luclin, that still only added a couple more decent dungeons for high level players. There were a lot more high level zones, but you weren't really going to go to Western Wastes or Katta Castellum to grind. I'm sure a few people did but it wasn't a first or even second/third choice for the vast majority of players for good reason.

    If each region can support a few groups worth of players, then a zone the size of Wild's End should easily be able to support 100 people, but I'm very skeptical of that when it comes to dungeons. People look back and say "X dungoen could hold Y people!" Yes it could HOLD that many players but it couldn't actually support that many players. I just think of the goblins in High Keep. Really, that's only 1 or 2 groups worth of goblins, but instead you'd have a group in each of the 4 rooms spinning their wheels between spawns. There's a big difference between "well that's all of the mobs, we got like 3 minutes before the next respawn. Smoke 'em if you got 'em." And "Well it took us 3 mintues to get through all of the mobs, now you can AFK for 20 minutes before the next respawn."

    Hopefully with so many pledgers being veteran players, we can organize zone raids where we try and fit as many groups as possible into places like Halnir's or Amberfaet and see how well the zone really can SUPPORT X number of players in a "live" situation.

    • 2752 posts
    September 15, 2021 11:28 AM PDT

    Really doesn't seem like something they can really get a good idea about until the world is actually built and tested, especially when it comes to how many total players per server. I definitely agree that higher mob density would go a long way to allowing far more players to co-exist nicely. I think having much higher population caps for servers would be ideal here so players don't run into many issues finding groups. 

    • 1921 posts
    September 15, 2021 2:23 PM PDT

    IMO:

    That's the exact opposite of how it should be done, Iksar.  There's 22+ years of experience in the genre that specifically offers guidelines regarding what is an ideal server population.
    That value; server population target/cap, is a fundamental design decision which has impacts on many/some/all other design decisions, in context.
    Ideally, you set the server pop goal range, then you design, then you build.  Otherwise, it will have almost no chance of being fun, challenging, or thematically/internally consistent.

    The whole design process must have some of these core or critical design decisions made at the start.  Otherwise, you're gonna have a bad time. :)  And repeat history, in a profoundly terrible way.

    • 2752 posts
    September 15, 2021 2:55 PM PDT

    vjek said:

    IMO:

    That's the exact opposite of how it should be done, Iksar.  There's 22+ years of experience in the genre that specifically offers guidelines regarding what is an ideal server population.
    That value; server population target/cap, is a fundamental design decision which has impacts on many/some/all other design decisions, in context.
    Ideally, you set the server pop goal range, then you design, then you build.  Otherwise, it will have almost no chance of being fun, challenging, or thematically/internally consistent.

    The whole design process must have some of these core or critical design decisions made at the start.  Otherwise, you're gonna have a bad time. :)  And repeat history, in a profoundly terrible way.

    22+ years of experience that can only offer guidelines as to what an ideal server population is for whatever the individual given world and it's content + mechanics are. Less so for many MMOs in that timeframe as their "server" populations have little impact on most activities as they are actually server clusters and you can play/group with people from any number of other servers. 

    I don't see why it would be important to nail down specific server population numbers before building out the world. Seems more like something done as the world is being built and they get a general idea of roughly how many players each zone can handle of any given level ranges, and as all zones eventually come in they can really settle on based on what "feels" right for both high and low population when it comes to content availability and ease of finding others to group with in different level ranges. Seems like a more fluid thing. As long as the world can hold at least thousands of players (which they'd be able to tell long before finishing the whole world), they can stay fluid with end population caps and content designs as they narrow in toward world completion/testing. 

    • 1921 posts
    September 15, 2021 5:21 PM PDT

    IMO:

    Because if you don't set the goal, you end up with EQ1.  It shouldn't be a fluid, unknowable, nebulous target, goal or thing.

    And.. "at least thousands" ? that's more than what they've specified, in over 7 years.  As a result we have no idea if that's accurate or not, objectively.

    • 9115 posts
    September 16, 2021 4:11 AM PDT

    This topic has been promoted for my CM content, please continue the discussion and have fun! :)

    "Hot Topic - Player To Zone Ratios: An exercise - An interesting discussion started by one of our community members, Benonai, on the official forums, what are your thoughts? https://bit.ly/3ltyCKC #MMORPG #CommunityMatters"

    • 32 posts
    September 16, 2021 4:34 AM PDT

    I don't buy the argument you can't have a community if there are more than X players on a server. There are billions of people on this earth and millions of communities. Make the servers as lively and as full as possible, and people will form tons of communities on the server. We don't need to all know each other. We don't need to all know of, or even of heard of the top guild, top player, etc, to have a community. We just need to be dense enough to have more people around us than we can know. We need the group of people we know to be just a small part of who all is running around. That will make it feel like a lived in world.

    • 2419 posts
    September 16, 2021 7:06 AM PDT

    hushed said:

    I don't buy the argument you can't have a community if there are more than X players on a server. There are billions of people on this earth and millions of communities. Make the servers as lively and as full as possible, and people will form tons of communities on the server. We don't need to all know each other. We don't need to all know of, or even of heard of the top guild, top player, etc, to have a community. We just need to be dense enough to have more people around us than we can know. We need the group of people we know to be just a small part of who all is running around. That will make it feel like a lived in world.

    It isn't that you can't have a community if more than X people are on it. Its more about the civility of the server can, and does, drop dramatically if there are far more people on a server than the content can support at any one time.

    • 337 posts
    September 16, 2021 8:00 AM PDT

    I feel like Vandraad has brought me back full circle on this one.

    I believe, not know, that VR DOES have a number that they are shooting for.  If you look at first build of AVP vs the new build of AVP (I'm not a VIP so this is anecdotal or based on what is known community-wide), they did not change the main hubs of content, just the zone size to fit the content.  That tells me that they have an amount of player activity they are looking at hosting on each zone then building out the world to suit it.

    If that's true, the fact that they released a map of the entire continent of Kingsreach would tell me that they have roughed out, by Pre-Alpha testing, what size locations they need to support x number of players with the amount of content they intend to have to satisfy the x.

    It's not lost on VR. I've heard it said a hundred times now.  Content is king.  The main attractions planned for Wild's End haven't changed in 5 years; the same ones they showed and discussed in original lore and in early Cohh streams back in 2016/2017. Formula for VR:

    Content (congruent to) # of players (congruent to) world size

    It appears to have started with content and ended with world size.  Obviously, that's not entirely true, but it's not that far off. That is why in original post, using world size to postulate how many players VR is intending to put on a server makes sense to me, at least.

    • 2138 posts
    September 16, 2021 8:26 AM PDT

    <- not servernuts-and-bolts/load engineering knowledgable

    How many PC's/NPC's, and where it might be important?: Starting Cities. In a new game thats where everyone is going to be and flooding into during the first month. In a desired crushingly slow exping game where half of max level is achieved in 2 years on "normal" casual play (2hrs, 3 nights a week and 6 hrs a day on weekends) ? The starting cities will be where its at, in fact, the perception may be that the starting city/zone is the "game" considering the level range of mobs, the vast areas to explore and do things in, and the amount of people in it. This is where zone design is crucial per NPC level/range. (all level 1-3 NPC's along the edge of the paths?, some around campfires inland a bit?- that way all the newbies can find something even if the whole place is full? 100 newbies in groups dotted along the zones pathways killing rats here, bandits there, bugs over there.) ? 

    Thats where I think the play vs resourse consumption test will be, especially if the cities will be dynamic and higher level players will find a reason to come back and interact.  Lets not forget the Whims of Devs that want to have events and suddenly the whole server is hell-bent on getting to Syronnai's rest to see whats up, all X00's of them getting to one zone.

    Who remembers Gfay in the early days? or ak anon, or kaladim?- packed! at least you guys could travel to each others places somewhat safely- they were close. I was in erudin that was a ghost town. I was the queen! lol. Even qeynos was a bit of a backwater, some creepy RL dudes. But some barbarians were cool (hi Calliach!, Lessa!) someone escorted me through BB to Halas and it was a blast, no one wanted to leave halas, I didnt want to leave halas lol. Lots of barbarians there doing stuff, different levels of mobs, plenty of room and ratio of spawns, could try something hard if you wanted. No Malachite though back in those days ,I had to ask higher levels passing through for special orders, if they could remember to remember me, some did.

    • 3852 posts
    September 16, 2021 8:29 AM PDT

    >I don't buy the argument you can't have a community if there are more than X players on a server.<

     

    Of course you can, and will. It just may not be as *good* a community. Over many years I have seen few exceptions to the rule that the civility of a community is inversely proportional to the population of the server. Other things being equal, of course (same game - same geopraphic area). 

    • 273 posts
    September 16, 2021 1:51 PM PDT

    hushed said:There are billions of people on this earth and millions of communities.

    That's debatable. There's a strong argument to be made the modern mass-society has displaced real community interaction (where people function together and account for their own actions), in favor of shallow technocratic relationships.

    MMOs, as a microcosm of real society, aren't all that different in how they've progressed in design.

    • 888 posts
    September 16, 2021 8:45 PM PDT
    I prefer higher populations because it's easier to find teams and the cities feel active. If we have large zones, there should be several places that feel empty, which is also important to me because it helps make those areas feel remote and help me feel like an explorer and not just a tourist.

    Ideally, there will be a way to dynamically scale so that low player population times and high population times feel similar.