EVE Search - Playable Limits and the Future

All Channels

EVE General Discussion

Playable Limits and the Future

» Click here to find additional results for this topic using Google

Monitor this thread via RSS [?]

Pages: 1 2 :: [one page]
Author	Thread Statistics \| Show CCP posts - 1 post(s)
Verus Potestas Caldari Fiat Mort	Posted - 2006.10.04 21:10:00 - [1] I was just thinking, do you think Eve will ever cap out at a maximum playable PCU, or will the server keep growing and getting expanded? Since I joined, the player base has tripled, the server has been upgraded with a sexier one, and we've been through countless optimisations. However, as it is, many things are laggier than they were when I joined, simply through sheer volume of players. The EVE player base is still growing fast (and accelerating in fact), and even more players will probably be drawn in by Revelations (and it's accompanying advertising). Is there going to be a limit where it will simply have to cap out, or can we keep growing forever, upgrading as people come along? What about fleet battles? More players than ever before are involved in each fleet battle, with some quite problematic results (Node> gurk). We have stress testing going on at the moment to try and help these things. What's the plan? RAWR!111 Sig Hijackz0r!!11 - Immy
Locke DieDrake Port Royal Independent Kontractors Imperial Republic Of the North	Posted - 2006.10.04 21:21:00 - [2] Is there a limit? Yes. Absolutely. BUT, assuming both further optimization and further hardware upgrades (more nodes, with more power each), then we are a long way from that limit. If you wanted to, you could build a 5000 node Blade server based on a beowulf cluster and cross link it with fiberchannel and run each and every solar system on a quad processor quad core blade of it's very own. Given that, and a database server to keep up, we are a long way from "caping" the PCU. Also Moores law is in affect. (this assumes that adding players equals a linear progression in usage of resources on the server) ___________________________________________ The deeper you stick it in your vein, the deeper the thoughts there's no more pain. ___________________________________________

kieron	Posted - 2006.10.04 21:22:00 - [3] We have hardware capable of handling a higher PCU than we have had recently, but Tranquility's stability has been causing more than its share of issue. Tranquility will be receiving a new RAMSAN in the near future and we're not going to stop there. The server architecture is such that we can continue to scale upward as the player base and PCU increases for quite some time. As technology improves, so will our hardware. We're not planning on splitting the EVE community across multiple servers any time in the foreseeable future. kieron Community Manager, EVE Online

Xenofur Di-Tron Heavy Industries Knights Of the Southerncross	Posted - 2006.10.04 21:32:00 - [4] kieron, out of curiosity: how will your hardware be able to hold up against fleet battles with 1000+ participants?
St Dragon Blood Association of Dragons	Posted - 2006.10.04 21:34:00 - [5] Originally by: Xenofur kieron, out of curiosity: how will your hardware be able to hold up against fleet battles with 1000+ participants? It would be a bit like this. KA-BBBBOOOOOOMMMMMM ----------------------------------------------- "Kill one man, and you are a murderer. Kill millions of men, and you are a conqueror. Kill them all, and you are a god." -- Jean Rostand
Rod Blaine Evolution Band of Brothers	Posted - 2006.10.04 21:35:00 - [6] Originally by: Xenofur kieron, out of curiosity: how will your hardware be able to hold up against fleet battles with 1000+ participants? Just like it does now, it won't. Because scaling the hardware (and software) up based on subscription count and pcu count (as it seems CCP have been doing so far) always has left them unable to deal with the largest battles the playerbase could come up with. Of course, that'd be different if some type of breakthrough is found that applies to large amounts of sessions changes at the same time and to large amounts of active characters in one grid specifically. Old blog
Miss Overlord Gallente BongBrothers Inc.	Posted - 2006.10.04 21:40:00 - [7] so its either a new software line or some new hardware that mgiht not be up straight powerful but could have new features that allow more acitivity within a node
Jenny Spitfire Caldari	Posted - 2006.10.04 21:55:00 - [8] Originally by: Xenofur kieron, out of curiosity: how will your hardware be able to hold up against fleet battles with 1000+ participants? You will need a dedicated node for that. --------- Cruelty is God's way of showing kindness and God is kind.
Quutar Caldari Uxor Infensus	Posted - 2006.10.04 21:59:00 - [9] my understanding right now is that: 1 node = 1 physical server in a cluster 1 node can hold multiple solar systems, but a solar system can only be on a single node. Every down time they run a load balancing script that looks at the load that the various solor system encountered and shifts the allocations of the nodes vs solor systems around to try to spread out the processing, and to give really busy systems thier own node, for instance jita is on it's own node. To move a solar system from one node to another while currently love will dis connect everybody currently on the solar system, and has potential bugs involved with POSes. so... ultimatly a single server will handle a single solar system. At some point it always has to boil down to a single server. but where is the bottle neck. when a node fails... why does it fail? is it due to lack of cpu, lack of memory, or lack of back end system (aka database). I am assuming that the database is scalable, and is a MS-SQL server (I am familiar with oracle... and I know we can cluster them on massive 300+ cpu machines if needed). (if the node death is due to the database... then they need to contract with Ms profesional services and with HP to get super DBs) Lets assume that the node fails due to cpu or memory. Both of these have a theoretical limit that a single server can hold, expecially if these are windows machines (I am assuming that they are, since CCP is a microsoft development partner). So just throwing more machines at the problem will not solve it, since the 800 man battle is at one of the machines. So what can they do. anything they can do to fix it will be a fundimental change to the EVE-Online client. I think the only thing they can do is to break down the "partitioning" further. Instead of limiting one system to a node, they limit one "grid" to a node. I think that this be a massive undertaking, if they did it, and would basically compleatly rewrite the server and network portion, maybe even the client portion of the game. Maybe there is something else, and since I don't have code access to thier apps, I can't make any realistic observations, other than wild ass guesses. I wonder if CCP has considered getting some loadrunner experts on staff, then they can simulate thousands apon thousands of clients in a controlled fashion, to find the problems and bottlenecks. Meh... it's a thought. sonofabeachballbouncingmarymotherfiretrucker
Forsch Auctoritan Syndicate Curatores Veritatis Alliance	Posted - 2006.10.04 22:02:00 - [10] Originally by: Xenofur kieron, out of curiosity: how will your hardware be able to hold up against fleet battles with 1000+ participants? That's why we got doomsday devices. Forsch Defender of the empire More love for side factions!
Admai Sket	Posted - 2006.10.04 22:07:00 - [11] I once heard that computing power for consumers doubles every year. I remember once that I got a PC which had 128MB ram and that was really living - I now have 2GB, and that is fairly decent at the moment i suppose. I dare say that after the new year, 4GB will be the norm. Unless the player base is doubling every year, I don't think we have too much to worry about.
evistin Multiverse Corporation	Posted - 2006.10.04 22:16:00 - [12] Moore's Law states that CPU power doubles, but no company I know of upgrades their computer every 18 years. Its just too expensive. The hardware CCP got now, are Duel core systems. They may buy a few Quad cores( assuming the software is scalable), but I think they will another generation or 2 before another upgrade. ----------- Management and Leadership û The Eve-online Guide
Fat Willy	Posted - 2006.10.04 22:16:00 - [13] Simple, we'll have 100,000 players online, but it will take 10 minutes to undock from a station, etc. When eve started up, undocking was virtually instantaneous. Now there's a built in delay to slow down the hit-rate on the server, to cope with the bigger player base.
Admai Sket	Posted - 2006.10.04 22:21:00 - [14] Originally by: evistin 18 years you do mean MONTHS, right?
Eilie Minmatar	Posted - 2006.10.04 22:37:00 - [15] Originally by: kieron We have hardware capable of handling a higher PCU than we have had recently, but Tranquility's stability has been causing more than its share of issue. Tranquility will be receiving a new RAMSAN in the near future and we're not going to stop there. The server architecture is such that we can continue to scale upward as the player base and PCU increases for quite some time. As technology improves, so will our hardware. We're not planning on splitting the EVE community across multiple servers any time in the ~~foreseeable~~ future. Fixed. Adding another server will be the death of EVE.
Spaja Saist Gallente Isotope Industries	Posted - 2006.10.04 22:44:00 - [16] Originally by: Verus Potestas I was just thinking, do you think Eve will ever cap out at a maximum playable PCU, or will the server keep growing and getting expanded? Since I joined, the player base has tripled, the server has been upgraded with a sexier one, and we've been through countless optimisations. However, as it is, many things are laggier than they were when I joined, simply through sheer volume of players. The EVE player base is still growing fast (and accelerating in fact), and even more players will probably be drawn in by Revelations (and it's accompanying advertising). Is there going to be a limit where it will simply have to cap out, or can we keep growing forever, upgrading as people come along? What about fleet battles? More players than ever before are involved in each fleet battle, with some quite problematic results (Node> gurk). We have stress testing going on at the moment to try and help these things. What's the plan? Maybe you haven't noticed but it seems Eve has already reached it's maximum playable size. They should of suspended trial accounts until they fix all of the problems Tranq is having. Well to be honest they should of stopped with the trial accounts to stop all the isk farmers.
Darksaber64x Ecchi co.	Posted - 2006.10.04 22:45:00 - [17] I don't think eve will ever cap out. Reasons I think this: 1) CCP will upgrade their hardware/software on regular intervals in order to maintain some server stability. 2) There is (sadly) a limited market for this game. Many people are offset by the sandbox nature, that you can do anything. Many people would rather have what they need to do laid out in front of them, mindlessly doing what they need to do to advance. (I.E. Grinding) 3) You can actually lose stuff. This off-sets even more people. They lose their ship, whine about it, petition it, and leave when they realize they can't get it back. 4) This game requires thinking. Most of the population isn't good at this, so will get frustrated and leave. So really the only cap we'll reach is the one where everyone that would be interested in eve are playing. Which I doubt will ever reach even 1 million (although that would be interesting having one million people on TQ (if/when it can handle that many people)).
Sadistic	Posted - 2006.10.04 22:48:00 - [18] I think Moore's law was something like "the number of transitors on a chip doubles every 2 years". Which leads to a chip being possible twice as powerful every two years. Its not a law (as in physics), it was an observation made in the 1960's that turns out to have been close to true for all this time. The earlier poster had it correct though. A dynamic (within 30 seconds during operating times) relocation of resources is the only solution. Eventually the problems we see in fleet battles (a large number of people suddenly showing up on a grid/system where there were previously few) will start to happen outside of fleet battles. Since the server optimizations happen during downtime, they can't handle this sudden change. So when that industrial corp brings it's 400 miners to that seldom used 1.0 sytems, it will have the same lag we see now in fleet battles (we are talking about the future with 100k pcu). CCP will have no choice but to fix the reallocation method. I would imagine CCP already see's this as the future and is working towards a solution. Whether they fix it by rewriting the engine, or through virtualization (or any other method), does not matter.
evistin Multiverse Corporation	Posted - 2006.10.04 22:48:00 - [19] Originally by: Admai Sket Originally by: evistin 18 years you do mean MONTHS, right? Ops my bad ----------- Management and Leadership û The Eve-online Guide
evistin Multiverse Corporation	Posted - 2006.10.04 22:55:00 - [20] Edited by: evistin on 04/10/2006 22:55:39 30 sec? It makes you wonder how much of the overhead will go into just managing the allocations then. Granted it has to be done, but it will be all but impossible to be this dynamic. Werid idea, All eve clients are roped in to provide addtional CPU to non-urgent tasks? Or a PS3 to run Jita anyone? lol ----------- Management and Leadership û The Eve-online Guide
Mak'shar Karrde Minmatar UK Corp Lotka Volterra	Posted - 2006.10.04 22:59:00 - [21] The way I understand it, you can just continue adding hardware as player numbers increase but only so long as they stay spread out. It's when players group up in large numbers on the same node that we start to have problems. I don't see how that problem can be so easily waved away.
Locke DieDrake Port Royal Independent Kontractors Imperial Republic Of the North	Posted - 2006.10.04 23:00:00 - [22] Originally by: evistin Moore's Law states that CPU power doubles, but no company I know of upgrades their computer every 18 years. Its just too expensive. The hardware CCP got now, are Duel core systems. They may buy a few Quad cores( assuming the software is scalable), but I think they will another generation or 2 before another upgrade. You fail to understand the scalability of the eve cluster. When they did the last set of upgrades, they made a choice to use IBM BLADES. They used blades because you don't have to "replace" the system every time you upgrade. You can litterally just buy some new pieces and throw it into the exsisting system. Last I heard there were only 72 nodes (blades) on the current cluster. Presumabely, the limit is infinite (actually, it's probably a multiple of 256 -1, but lets not go there) No one from CCP has said how many blade centers (the box that holds and networks the blades together) they have. But this can also just be tacked onto the exsisting cluster at any time. It's obvious by now that the software scales to the number of nodes. Not the other way around. So we have 5000 solar systems on 72 nodes. Double the number of nodes (which is easy in the current architecture, if not exactly cheap) and you've affectively doubled the PCU that the cluster can handle. This however doesn't address the issue of large scale battles... that I'm not going to get into here. In the most basic sense... CCP can add nodes whenever they have money, and new players is new money . Even a massive influx of players can be scaled, given a certain amount of lead time. (of course, that lead time means days or weeks of node crashes while the new hardware is ordered and installed) And just for the record, I upgrade my 3 personal computers at least once every 12 months. It's uaually my birth day gift to myself. Most companies don't do this, but most companies are using the same software they were using 2 years ago, and don't have any need for faster desktop machines. Servers are a different matter all together. And are very often only in service for a matter of months. Like 6 to 12. My company buys a new NAS (network attached storage) every 4 months. Keeping the old ones and adding the new ones. (4petabytes and counting woot) And a new host server every 1 year or so. Replacing the old one which is usually downgraded to a minor function on the network (like cold storage or WIKI host). So yeah... 18 months is a LONG life for a computer. ___________________________________________ The deeper you stick it in your vein, the deeper the thoughts there's no more pain. ___________________________________________
Locke DieDrake Port Royal Independent Kontractors Imperial Republic Of the North	Posted - 2006.10.04 23:05:00 - [23] Originally by: Mak'shar Karrde The way I understand it, you can just continue adding hardware as player numbers increase but only so long as they stay spread out. It's when players group up in large numbers on the same node that we start to have problems. I don't see how that problem can be so easily waved away. This is correct based on my understand too. The issue is that single solar systems can only be hosted on single nodes. Therefore the amount of processing power that node has is DIRECTLY related to the number of people that can go there and not get problems. Certain systems are known to have their very own node. Jita for instance. But even jita gets laggy with 800+ people. And AFAIK no one has ever started a massive war there. In other systems, where the node is shared by many other systems.. the limit on people is much lower per system because they have to share. So fleet battles are ALWAYS going to be laggy based on the architecture of the server. It would be interesting to know if you could have a 500v500 fight on a single unshared node without major lag. If thats true, then we just need more nodes. If you can't have a big fight, even on a single unshared node, then you aren't ever going to be able to. CCP isn't likely to tell us this if it's the case, because alot of us alliance types play with the dream of having those big fights one day. ___________________________________________ The deeper you stick it in your vein, the deeper the thoughts there's no more pain. ___________________________________________
evistin Multiverse Corporation	Posted - 2006.10.04 23:08:00 - [24] Originally by: Locke DieDrake Stuff.. I see what you mean. I am aware that as a cluster they can add/remove nodes as they wish of varying hardware levels. Sadly I can't see an easy way to fix this problem. Maybe the new regions will help spread the player base out more, thus reducing the frequency of such events. ----------- Management and Leadership û The Eve-online Guide
Locke DieDrake Port Royal Independent Kontractors Imperial Republic Of the North	Posted - 2006.10.04 23:13:00 - [25] Originally by: evistin Originally by: Locke DieDrake Stuff.. I see what you mean. I am aware that as a cluster they can add/remove nodes as they wish of varying hardware levels. Sadly I can't see an easy way to fix this problem. Maybe the new regions will help spread the player base out more, thus reducing the frequency of such events. Best bet is for CPU power to increase faster. Which is likely to happen in the near future. Intel is talking about 80 core CPU's by 2008. Thats a big jump on the standard moore's law. There are also countless factors that we as players just aren't privy to. So it's possible that some factor the player base is just not aware of could change and reduce lag by 10000%. For all we know, deleting every bookmark from the game would reduce overall CPU usage by 75%. That would be a huge differance. ___________________________________________ The deeper you stick it in your vein, the deeper the thoughts there's no more pain. ___________________________________________
Sadistic	Posted - 2006.10.04 23:14:00 - [26] Anything less then 30 seconds (fleet A jumps into system to find fleet B camping the gate) would not help much. It would not be that hard conceptually. A system with limited resources (not one of the hardcoded systems like Jita) sees its getting busy, or that that it has lots of people on grid with the gate to another busy system). It gets mirrored to a dedicated box. During the mirror process they are duplicating each other. If something happens (the fleet jumps in), the dedicated mirror becomes primary and the non dedicated box is dropped from the process. After the event drops below a certain level (one fleet no longer exists), the data is once again mirrored until they are in sync, then the dedicated box drops out and becomes available for some other action. This still faces the limit of one system per node, but allows there to be a few mega nodes that will hopefully be able to handle it the sudden events anywhere in the Eve universe. On the other hand, since I have no idea how the Eve engine really works, maybe they could use my old Atari 800.
evistin Multiverse Corporation	Posted - 2006.10.04 23:22:00 - [27] We can't keep throwing more hardware at the problem. In the end, we will hit the end of the curve. At some point the software needs to accomdate the ability to manage such huge battles. I would be curious to see why we can't move players from Node to Node while playing. Moore Law is techincally still in effect when you see the roadmap intel put out. The law is specfically to transistor counts, the 80 core unit will share a fair bit of transistors to achieve its design. Oh I will make that bet wirh you that killing bookmarks will increase performance insanely. ----------- Management and Leadership û The Eve-online Guide
Keta Min Pre-nerfed Tactics Blood of the Innocents	Posted - 2006.10.04 23:24:00 - [28] Originally by: evistin Originally by: Locke DieDrake Stuff.. I see what you mean. I am aware that as a cluster they can add/remove nodes as they wish of varying hardware levels. Sadly I can't see an easy way to fix this problem. Maybe the new regions will help spread the player base out more, thus reducing the frequency of such events. the fix would be to shape game mechanics in a way that blobbing in one grid is not necessary and is also discouraged. this would improve gameplay aswell imo. unfortunately the POS sovereignty and POS/capital warfare mechanics are heading in the opposite direction at the moment.
Dragon	Posted - 2006.10.04 23:33:00 - [29] Originally by: evistin We can't keep throwing more hardware at the problem. In the end, we will hit the end of the curve. At some point the software needs to accomdate the ability to manage such huge battles. And when that happens the blobs will grow even larger. Fleet battles have always put TQ to the test and the end result has always been a lagfest. Back in the days 150-200 was the limit for a fleet battle and then everything started to lag up. With the improvements to the hardware that number has been boosted somewhat. The end result is still always the same. It's a vicious circle that will never end basically. The limit that CCP put on heavily loaded systems recently is actually one of the best ideas they've ever had to address this problem. Perhaps we'll actually get to participate in lagfree fleet battle someday.
Malthros Zenobia Caldari Independent Navy Reserve Kimotoro Directive	Posted - 2006.10.04 23:52:00 - [30] To answer the question of 'when' CCP will beable to make the server handle a fleet battle: Never. If tomorrow you could have a 400 person, lag-free battle, both sides would most likely bring 300+, and you'd have a lag-tastic 600+ person battle. IF in a year they can handle 1k people in a fleet fight, you'll have people trying to bring 1500 people to fleet fights. Originally by: kieron The Carrier was never intended to be a solo OMGWTF mission-farming PWNmobile.
evistin Multiverse Corporation	Posted - 2006.10.04 23:59:00 - [31] Maybe we should have a treaty, where all alliances who sign it, will not engage a enemy system or defend a system with more than 50 ships...lol. Like that will ever happen. ----------- Management and Leadership û The Eve-online Guide
w0rmy Intensive CareBearz	Posted - 2006.10.05 00:50:00 - [32] Originally by: St Dragon Originally by: Xenofur kieron, out of curiosity: how will your hardware be able to hold up against fleet battles with 10+ participants? It would be a bit like this. KA-BBBBOOOOOOMMMMMM There, fixed Originally by: Avon Originally by: Dark Shikari What single item is larger than a jetcan? My ego?
Hllaxiu Shiva Morsus Mihi	Posted - 2006.10.05 01:05:00 - [33] Originally by: Locke DieDrake Best bet is for CPU power to increase faster. Which is likely to happen in the near future. Intel is talking about 80 core CPU's by 2008. The question is how many can the server side software use. 80 cores is not going to improve performance 80 times... --- Our greatest glory is not in never failing, but in rising up every time we fail. - Emerson
Locke DieDrake Port Royal Independent Kontractors Imperial Republic Of the North	Posted - 2006.10.05 01:50:00 - [34] Originally by: Hllaxiu Originally by: Locke DieDrake Best bet is for CPU power to increase faster. Which is likely to happen in the near future. Intel is talking about 80 core CPU's by 2008. The question is how many can the server side software use. 80 cores is not going to improve performance 80 times... Well no. But dual cores, on unoptimized systems usually provide about a 50% increase in CPU overhead. Which means effectivly taking one current node and making it 40x more powerfull. (using an 80 core processor) (this also presumes that the clock speed doesn't change, and also that memory wont' improve, which both will) I'm not sure it really works like that. There is probably a ceiling somewhere in there. Memory bus or execution block size or some other detail. But who knows. I'd presume that EVE is optimized to run on multi-processor systems to begin with. But I could be off there. If it's not, then that is going to have to change in order for eve to take advantage of the massive scaling of cores that we are going to see over the next few years. ___________________________________________ The deeper you stick it in your vein, the deeper the thoughts there's no more pain. ___________________________________________
Seventh Paradox Gallente	Posted - 2006.10.05 03:31:00 - [35] The hardware is not the limiting factor in the way eve works. It's the architecture used to achieve it. You can upgrade the hardware every few months but it won't necessarily mean a performance improvement in real terms. There's really nothing to fear from CCP maxing out any of their hardware capabilities if they plan accordingly with their architecture. ItÆs poorly written code and clunky software layers that cause the problems. I work for one of the largest financial transaction companies in the world designing and supporting their server infrastructure. Oracle rates us as one of the hottest 3 databases in the world. (You're my neighbours by the way CCP, we rent some of that same Level 3 complex) We use multiple SUN 6900 DB servers (The enterprise versions have separate memory boards that result in adding a couple of hundred nanoseconds to latency which affects performance to much.) with 48 CPU's, 48GB of RAM and FibreChannel connectivity to a 22TB HDS. We push over 13,000 hits per second to the web and a constant 600MB/s of outbound web traffic without accounting for any peak time. We usually run at a load of around 4.0 (The equivalent of 4 CPUÆs having instructions in their queue.) Our web front end consists of 60 Sun X4100 boxes running Linux that distribute our services.Our cluster barely breaks 4% usage with normal load. So there is no hardware limit that CCP will reach in the next few years that existing hardware can't manage. It's whether or not the software architecture is scalable enough to handle that many transactions. So all we can do is speculate which isn't very productive unless one of you can show me a flowchart of the software tier and the network infrastructure. So keep playing, keep paying and all the back end stuff will take care of itself.
evistin Multiverse Corporation	Posted - 2006.10.05 03:53:00 - [36] Is that not what CCP is doing now, working to optimize the code, part of the recent patch/Dragon code was to improve server performance and stablility. ----------- Management and Leadership û The Eve-online Guide

Pages: 1 2 :: [one page]

COPYRIGHT NOTICE
EVE Online, the EVE logo, EVE and all associated logos and designs are the intellectual property of CCP hf. All artwork, screenshots, characters, vehicles, storylines, world facts or other recognizable features of the intellectual property relating to these trademarks are likewise the intellectual property of CCP hf. EVE Online and the EVE logo are the registered trademarks of CCP hf. All rights are reserved worldwide. All other trademarks are the property of their respective owners. CCP hf. has granted permission to EVE-Search.com to use EVE Online and all associated logos and designs for promotional and information purposes on its website but does not endorse, and is not in any way affiliated with, EVE-Search.com. CCP is in no way responsible for the content on or functioning of this website, nor can it be liable for any damage arising from the use of this website.