l3al3e Posted November 22, 2012 Share Posted November 22, 2012 a server that is equivalent to 4 sims? Like recreational sims. Sometimes I want to go really fast in my boat but needs to slow down to cross over into the next sim just in case the next sim isn't ready for me yet. Would be nice to go fast long distance with out the hick up. Link to comment Share on other sites More sharing options...
Amethyst Jetaime Posted November 22, 2012 Share Posted November 22, 2012 I'm sure they could do this however from an economic point of view I don't think there would be much, if any, demand for them. A full sim is 295US a month. A sim the size of four regular sims would be way out of the price range of just about everyone. Even if they didn't support as many prims as a full sim, just the equivalent of a homestead x 4, they still would cost a lot more than a regular full sim. With the exception of welcome centers, info hubs a few sailing regions and sandboxes, all the land in SL is either owned by a private party or will be sold to one at some point. LL's business model is to be paid for as much land as possible and to charge what the market will bear for it. They aren't about to supply megaregions for free for people to use. It wouldn't be just the cost of the servers but also development costs as the architecture of SL is set up for 256sqm regions only. This has been brought up a number of times over the years as well as people asking about private islands half the normal size with half the prims. I'm sure if they saw a real demand for it and it was cost effective for them we'd have them. Link to comment Share on other sites More sharing options...
l3al3e Posted November 22, 2012 Author Share Posted November 22, 2012 I was thinking of sims owned by LL like parts of the Blake Sea or other large bodies of water that they own or even where they have that track near Morris. Right it would be really expensive to own all that amount of land. Link to comment Share on other sites More sharing options...
161488303349 Posted November 22, 2012 Share Posted November 22, 2012 is quite interesting this discussion when it comes up + just to raise the bar (: 4 sims is not enough for any kind of vehicle travel really. only goes from 256m to 512m in a straight line next square is 9 sims 3x3 for 768m straightline 4x4 is 16 sims for 1km 2km would need 8x8 64 sims altho could make strips. say 8 x 2 to reduce sims. or whichever and so on + ideally for big big oceans/voyages and avoid sim handoffs/crossings (and crasho) then have to have a avatar-centric simulator where the vehicle/avatars never leave the actual hardware hosted (or virtualised) simulation they logged into. means that the world/scenegraph has to move/flow past the avatar(s) is completely different model to SL this one day someone will make this tho i think. when they do then a sailing/boating/flight world will probably be the target market Link to comment Share on other sites More sharing options...
Phil Deakins Posted November 22, 2012 Share Posted November 22, 2012 That's how the old multi-user scrolling text games worked - and single-user adventure games. People didn't move to locations as they moved around the place. Instead they were simply provided with the data that was relevant to the location they 'appeared' to move to. As a programmer, it used to puzzle me until I decided to write one. Then the method became obvious. Using the same method, I also wrote a simple graphics one that I had online with multiple users, but only for testing and demonstrating. It could be done with a system like SL but it would be a completely different way of doing things. So different that it probably isn't worth LL doing it just for the sake of smooth passage around the various suitable areas like seas. I don't know though. Maybe it would be worth doing, and creating special areas for it. Link to comment Share on other sites More sharing options...
161488303349 Posted November 22, 2012 Share Posted November 22, 2012 yes. cant see linden doing it either really maybe will be in SL 2.0 if there is ever one + one thing they could do tho to make sim crossings less frustrating is make some client side change to menu: Develop \ Network \ Velocity Interpolate Objects at the moment if turn this off then you dont go random zooming off into space or whereever when is crossing lag when you do turn it off tho then you get stop/start stutter all the time when you move. as the viewer not do interpolation while waiting for server + but if the interpolation was turned off automatic clientside while a sim handoff was happening then vehicle/avatar will stop/pause until handoff is complete if we got a onscreen advice indicator (like the timers/debug advices) saying "pause waiting for sim handoff" and then start moving again after handoff complete then would be better i think better for me anways bc i rather have a stop/pause than watch my avatar just zoom off into wonderland and sometimes (like often) never recover and crasho could even put a time period on it somehow. like if handoff not complete in xtime? then is cancelled and the advice says "handoff fail" and you bounced back a wee bit (like how happens on a void sim boundary) so can either try again or turn and go another way Link to comment Share on other sites More sharing options...
Orca Flotta Posted November 23, 2012 Share Posted November 23, 2012 On OSG they are successfully running mega regions of 4, 9 and 16 sims already. Link to comment Share on other sites More sharing options...
Phil Deakins Posted November 23, 2012 Share Posted November 23, 2012 Explain please. Do you mean that they are running a single sim that measures 512m x 512m (that's the equivalent land area of 4 normal sims) and bigger? And, if they are doing that, do those megasims have normal sims on their borders? E.g. 2 normal sims along each side of a 512x512 sim. Link to comment Share on other sites More sharing options...
Orca Flotta Posted November 24, 2012 Share Posted November 24, 2012 I haven't got the faintest idea, Phil. I only read it in some forum or sumsuch that some developers (isn't everyone on OSG a developer?) did it and that it catches on now . My own experience with OSG is like 10 minutes when I dared to register for that grid sometime in 2008, and crashed out while trying on a freebie outfit. So in fact I'm an early adopter, a old salt in OSG ... but still a buttnaked n00b. :smileyembarrassed: Link to comment Share on other sites More sharing options...
Amethyst Jetaime Posted November 24, 2012 Share Posted November 24, 2012 From what I understand from talking to some open grid developers mega regions are not true mega region like the OP is asking about. Instead when you cross a sim line the server for the sim you were on, continues to provide your data until all of it has downloaded to the next sim server signals the first server that it will take over now. It works smoothly, but not always perfectly, when you walk from one sim to another but I have heard vehicle crossings also have to be slow for it to work. If you are going fast, you can be through a sim before the process is complete and have the same problems you would in SL. A number of the open grid worlds experimenting with mega regions don't have physics yet either or they are still in the early stage of developing their physics and so have no vehicles yet or the ride is even rough staying in one sim. Of course I have not visited every grid, only the more 'advanced' ones and some not open to the public, so maybe someone has worked something else out. I've never been to OpenSim because it is too "open' to be viable for business. Link to comment Share on other sites More sharing options...
Perrie Juran Posted November 24, 2012 Share Posted November 24, 2012 The issue is Scalability. While I don't understand all the technical aspects I think I understand the basic ideas here. http://www.cs.cornell.edu/people/~wmwhite/papers/2009-ICDE-Virtual-Worlds.pdf "Abstract—Networked virtual environments (net-VEs) are the next wave of digital entertainment, with Massively Multiplayer Online Games (MMOs) a very popular instance. Current MMO architectures are server-centric in that all game logic is executed at the servers of the company hosting the game. This architecture has lead to severe scalability problems, in particular since MMOs require realistic graphics and game physics – computationally expensive tasks that are currently computed centrally. We propose a distributed action based protocol for net-VEs that pushes most computation to the computers of the players and thereby achieves massive scalability. The key feature of our proposal is a novel distributed consistency model that allows us to explore the tradeoff between scalability, computational complexity at the server, and consistency. We investigate our model both theoretically and through a comprehensive experimental evaluation." Second Life, as explained in that article, has a very high computational complexity. It is simply it's nature, especially when compared to other MMO's. The data has to be broke down into manageable pieces relative to the available computational power available. Both Server and Client have to be taken into account. While it may be easy to say, "just make the SIM larger," all the data still has to be managed and processed. And while having fewer SIMs per server might possibly improve things Server side, there still will be the Client to deal with. A low end computer will still "lag" just as bad. Link to comment Share on other sites More sharing options...
Pussycat Catnap Posted November 24, 2012 Share Posted November 24, 2012 Phil Deakins wrote: As a programmer, it used to puzzle me until I decided to write one. Then the method became obvious. Using the same method, I also wrote a simple graphics one that I had online with multiple users, but only for testing and demonstrating. It could be done with a system like SL but it would be a completely different way of doing things. The entire structure of SL is badly designed... Its not ideal for a multi-user world on so many levels... Whoever came up with the underlying logic didn't really think it through. MMOs have shown that other ways of connecting everything together are much better - and if anyone ever takes some of that architecture and puts it to one of these graphical MUSHes, SL is dead. Link to comment Share on other sites More sharing options...
Perrie Juran Posted November 24, 2012 Share Posted November 24, 2012 Pussycat Catnap wrote: The entire structure of SL is badly designed... Its not ideal for a multi-user world on so many levels... Whoever came up with the underlying logic didn't really think it through. MMOs have shown that other ways of connecting everything together are much better - and if anyone ever takes some of that architecture and puts it to one of these graphical MUSHes, SL is dead. Do you actually understand what you are talking about or are you just parroting what you have heard others say? Read the article I linked above and then tell me that you think it is that simple. While I don't question that there are things that could be done better, I do question any one who thinks it's that simple. Link to comment Share on other sites More sharing options...
Pussycat Catnap Posted November 24, 2012 Share Posted November 24, 2012 I'm parroting myself. Simple - no. But SL is still put together all wrong at a core design level. Shouldn't even have a concept of a sim. Link to comment Share on other sites More sharing options...
Jadeclaw Denfu Posted November 24, 2012 Share Posted November 24, 2012 Seeing the massive parallel computers, that exist today, SL could be constructed completely 'simless', as all of the cpus can act a single multicore cpu, handling the whole world seamlessly. Plus, the servers process graphic related data, so it would profit from the new nVidia GPU-cores as well. However, back then the concept of dividing the whole world into squares was possibly the most logical way to start and upgrading it to the size we see today. When it comes to extending the world, the 'sim'-concept has its advantages: Sims (Servers) can be added and removed as necessary without affecting the rest of the world. Changing the concept today is practically impossible without breaking almost everything. Link to comment Share on other sites More sharing options...
161488303349 Posted November 25, 2012 Share Posted November 25, 2012 Pussycat Catnap wrote: Phil Deakins wrote: As a programmer, it used to puzzle me until I decided to write one. Then the method became obvious. Using the same method, I also wrote a simple graphics one that I had online with multiple users, but only for testing and demonstrating. It could be done with a system like SL but it would be a completely different way of doing things. The entire structure of SL is badly designed... Its not ideal for a multi-user world on so many levels... Whoever came up with the underlying logic didn't really think it through. yes they did really. think it through the way SL got designed was the only feasible way to do it on the hardware botn server and client side way back then. feasible meaning practicable and affordable even if seems a bit outdated now SL was engineering masterpiece for its time. like realtime dl of assets in a contiguous setting. in many ways is still really amazing that they even managed to make it + the simless/seamless world we talking about is not easy to do even now like how to move a world past a fixed point not just for one avatar but 100s or even 1000s all in the same space point makes the calculations required go up by orders of massive magnitude probably need at least one server core for each 4 or 8 avatars based on todays hardware. bc each avatar would be a sim all by themselves Link to comment Share on other sites More sharing options...
161488303349 Posted November 25, 2012 Share Posted November 25, 2012 Phil Deakins wrote: That's how the old multi-user scrolling text games worked - and single-user adventure games. People didn't move to locations as they moved around the place. Instead they were simply provided with the data that was relevant to the location they 'appeared' to move to. As a programmer, it used to puzzle me until I decided to write one. Then the method became obvious. Using the same method, I also wrote a simple graphics one that I had online with multiple users, but only for testing and demonstrating. It could be done with a system like SL but it would be a completely different way of doing things. So different that it probably isn't worth LL doing it just for the sake of smooth passage around the various suitable areas like seas. I don't know though. Maybe it would be worth doing, and creating special areas for it. the scrolling games usual only bi-directional. like up down or left right some move complex like they work from a limited number of fixed pov cameras. flat 2D or 2.5d pov compare to them tho SL way more complex (same other 3D game/worlds bc is 360 degree camera in 3D spacetime then is the raw power needed to do the calculations that hold them back at the moment Link to comment Share on other sites More sharing options...
161488303349 Posted November 25, 2012 Share Posted November 25, 2012 Perrie Juran wrote: The issue is Scalability. While I don't understand all the technical aspects I think I understand the basic ideas here. http://www.cs.cornell.edu/people/~wmwhite/papers/2009-ICDE-Virtual-Worlds.pdf "Abstract—Networked virtual environments (net-VEs) are the next wave of digital entertainment, with Massively Multiplayer Online Games (MMOs) a very popular instance. Current MMO architectures are server-centric in that all game logic is executed at the servers of the company hosting the game. This architecture has lead to severe scalability problems, in particular since MMOs require realistic graphics and game physics – computationally expensive tasks that are currently computed centrally. We propose a distributed action based protocol for net-VEs that pushes most computation to the computers of the players and thereby achieves massive scalability. The key feature of our proposal is a novel distributed consistency model that allows us to explore the tradeoff between scalability, computational complexity at the server, and consistency. We investigate our model both theoretically and through a comprehensive experimental evaluation." Second Life, as explained in that article, has a very high computational complexity. It is simply it's nature, especially when compared to other MMO's. The data has to be broke down into manageable pieces relative to the available computational power available. Both Server and Client have to be taken into account. While it may be easy to say, "just make the SIM larger," all the data still has to be managed and processed. And while having fewer SIMs per server might possibly improve things Server side, there still will be the Client to deal with. A low end computer will still "lag" just as bad. like it says in the paper one of the models is to move to more client side computation. like with the new gfx cards capable of doing quite advanced computations now the main hole in doing this has been the games that require combat and leveling. if to much control specially over movement and physics is depend largely on client side computations then the people with more powerful computers will always/mostly win. it also makes it easier to hax the game mechanics so can get big advantages over other players has been soemthing that lots of people over the years have tried to find a fix for. so far nobody really has solved it from engineering pov + in a environment where is no leveling or competitive game play then would be ok to do this way i think like the server side only just keep track of each character/avatar on the world scene. and handle world point assets ul/dl and comms and inventory and scripts (without the physics). everyhing else done client side mostly the physics so could then have like big oceans and waves clientside for them with computers that can handle it. like if is other person on your boat then on your screen you and them and the boat go up and down on the waves. and at same time is flat water for them without (say your passenger). so the boat for them just goes like in SL now Link to comment Share on other sites More sharing options...
Perrie Juran Posted November 25, 2012 Share Posted November 25, 2012 16 wrote: like it says in the paper one of the models is to move to more client side computation. like with the new gfx cards capable of doing quite advanced computations now the main hole in doing this has been the games that require combat and leveling. if to much control specially over movement and physics is depend largely on client side computations then the people with more powerful computers will always/mostly win. it also makes it easier to hax the game mechanics so can get big advantages over other players has been soemthing that lots of people over the years have tried to find a fix for. so far nobody really has solved it from engineering pov + in a environment where is no leveling or competitive game play then would be ok to do this way i think like the server side only just keep track of each character/avatar on the world scene. and handle world point assets ul/dl and comms and inventory and scripts (without the physics). everyhing else done client side mostly the physics so could then have like big oceans and waves clientside for them with computers that can handle it. like if is other person on your boat then on your screen you and them and the boat go up and down on the waves. and at same time is flat water for them without (say your passenger). so the boat for them just goes like in SL now This brings us back to one of the big differences between SL and MMO's. The Enviroment (World) in SL is dynamic, ever changing. In most MMO's it is static, it never changes. So at a minimum, everytime I enter a SIM in SL, even if it is cached on my Computer, a check has to be made if anything has changed on that SIM since the last time I was there. And then in real time, any changes being made......whether I add or delete a Prim or change a texture, that information has to be updated on the Server and also sent to any one on the SIM or who enters it later. Otherwise we lose the shared experience. How often have we experienced this: Last night I was out dancing with a friend. We both hopped on couples dance balls. I started dancing almost immediately. It was 30 seconds later before I saw them start to move. They even commented to me that it was taking a long time for them to start animating. It was a very crowded club, over forty people there, and they were on a lower end computer. Eventually I suggested that we go elsewhere to even out the shared experience. Link to comment Share on other sites More sharing options...
Echo Hermit Posted November 26, 2012 Share Posted November 26, 2012 Perrie Juran wrote: 16 wrote: like it says in the paper one of the models is to move to more client side computation. like with the new gfx cards capable of doing quite advanced computations now the main hole in doing this has been the games that require combat and leveling. if to much control specially over movement and physics is depend largely on client side computations then the people with more powerful computers will always/mostly win. it also makes it easier to hax the game mechanics so can get big advantages over other players has been soemthing that lots of people over the years have tried to find a fix for. so far nobody really has solved it from engineering pov + in a environment where is no leveling or competitive game play then would be ok to do this way i think like the server side only just keep track of each character/avatar on the world scene. and handle world point assets ul/dl and comms and inventory and scripts (without the physics). everyhing else done client side mostly the physics so could then have like big oceans and waves clientside for them with computers that can handle it. like if is other person on your boat then on your screen you and them and the boat go up and down on the waves. and at same time is flat water for them without (say your passenger). so the boat for them just goes like in SL now This brings us back to one of the big differences between SL and MMO's. The Enviroment (World) in SL is dynamic, ever changing. In most MMO's it is static, it never changes. So at a minimum, everytime I enter a SIM in SL, even if it is cached on my Computer, a check has to be made if anything has changed on that SIM since the last time I was there. And then in real time, any changes being made......whether I add or delete a Prim or change a texture, that information has to be updated on the Server and also sent to any one on the SIM or who enters it later. Otherwise we lose the shared experience. How often have we experienced this: Last night I was out dancing with a friend. We both hopped on couples dance balls. I started dancing almost immediately. It was 30 seconds later before I saw them start to move. They even commented to me that it was taking a long time for them to start animating. It was a very crowded club, over forty people there, and they were on a lower end computer. Eventually I suggested that we go elsewhere to even out the shared experience. I'm usually the one on the low-end computer that my friends need to take to a quieter venue :matte-motes-frown: I haven't anything technically intelligent to add to this thread; am really still impressed at how it does all work at all, when you consider behind the scenes it looks like this (bbang on first server) Governor Linden's mansion still has the old maps up, showing how much SL has grown over the years. Obviously a hell of a lot of thought was put into the original creation of Second Life, but now it is what it is. And I love it, warts, lag, funky sim crossings and all. :matte-motes-big-grin-wink: Might be fun, though, if LL let the regular residents who are also programmers in RL loose to make improvements (off to sew my sides back together again). Link to comment Share on other sites More sharing options...
Recommended Posts
Please take a moment to consider if this thread is worth bumping.
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now