EVE Search - So, what happens to the node after we send a large fight form?

Check My IP Information

All Channels

EVE General Discussion

So, what happens to the node after we send a large fight form?

» 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 - 3 post(s)
Wu Jiaqiu Minmatar Dreddit Test Alliance Please Ignore	Posted - 2010.12.13 04:18:00 - [1] Do they place a Jita strength server to back up the system to boost its speed?
dkbjitawhore	Posted - 2010.12.13 04:21:00 - [2] Edited by: dkbjitawhore on 13/12/2010 04:26:17 If you're lucky, yes. If your not lucky they simply make sure the designated system is on it's own server instead of having to share with other systems. For Example: Normally system A, B, C, D, and E (which lets say are a constelation) are one one server. If you anticipate there to be a big fight in system B and system D is a system the fight is likely to spill over into, you request reinforcement for system B and D and then both of those systems are re-mapped to their own server after DT until the next DT.
Lilla Kharn Amarr B4D W0LF Systematic-Chaos	Posted - 2010.12.13 04:25:00 - [3] No other system mimics Jita because Jita can always be expected to be laggy as all hell, therefore it always has several dedicated nodes running it. When you input a notice in, it lets CCP know that the node does need to be reinforced, but no other node will ever make it to where Jita is at today. ============================================= "Only the dead have seen the end of war" - Plato
JohnPaulJones Caldari DarkStar 1 Goonswarm Federation	Posted - 2010.12.13 05:44:00 - [4] Quote: So, what happens to the node after we send a large fight form? Well soon after a GM in an ibis will appear from nowhere and get podded by the already massed North Coalition and have to return in a ship we cannot kill and inform us he/she isnt happy we shot down and or podded him/her and quietly and hopefully reinforce the node so our module cycle drops from 3 hours to only 1 hour.
Akita T Caldari Caldari Navy Volunteer Task Force	Posted - 2010.12.13 07:48:00 - [5] Edited by: Akita T on 13/12/2010 07:54:24 Originally by: Lilla Kharn No other system mimics Jita because Jita can always be expected to be laggy as all hell, therefore it always has several dedicated nodes running it. When you input a notice in, it lets CCP know that the node does need to be reinforced, but no other node will ever make it to where Jita is at today. Wrong. First off, no, Jita does NOT run on "several" nodes. If that would be possible, they could have "negligible lag Jita" by tomorrow, and they could keep ahead of the lag-spike by adding more and more nodes to the mix. No system in EVE can run on more than one node, that is one problem CCP is trying (and not yet succeeding) to correct. It's harder than it sounds, thanks to how the initial code was written and the expectations people had for future CPU core speed advances (they weren't even considering the possibility multiple-core machines to become the norm, instead hoping for single core CPU speed increases). To pull that off, they'd basically have to gut and rebuild almost everything from scratch. And I'm not even sure we're only talking software here. Second, the hardware of the node Jita runs on is not unique, hasn't been unique for a while now, there are several of the same type, handling some other trade hubs AND used as reinforced locations for fleet fights that should be near-epic in scale, when available. So, yes, you COULD be having a fleet fight on hardware that's pretty much identical to the one Jita runs on. But that's still not going to help much if each side brings over a thousand ships to the party... Finally, you just need to make it known well in advance, because for now, a downtime is needed to reallocate a system to a different node. That's also one of the other things CCP is supposedly (trying to be) working on, the ability to switch between one node to another "on the fly". Which one of those things is harder to accomplish (on-the-fly relocation or multiple-nodes-per-system) is difficult to determine. There was a lot of talk about "Infiniband"//HPC on the hardware side, which was supposed to make both of those things possible or at least somewhat easier or better (you still need massive software rewrites either way)... but I don't even know if that's a big focus at CCP anymore or not. If it is, it certainly hasn't received much public attention lately. TL;DR - yes, you might get a Jita-like node all for the fight. Or you might get a "regular" node just for the fight. Or you might get a slightly less busy normal node for the fight. Or you might get nothing extra on top of the regular resource allocation for that system//node. All depends on what else might be happening and where, how early you made the request and so on. _ Beginner's ISK making guide \| Manufacturer's helper \| All about reacting _
Sturmwolke	Posted - 2010.12.13 10:13:00 - [6] Originally by: Akita T There was a lot of talk about "Infiniband"//HPC on the hardware side, which was supposed to make both of those things possible or at least somewhat easier or better (you still need massive software rewrites either way)... but I don't even know if that's a big focus at CCP anymore or not. If it is, it certainly hasn't received much public attention lately. There was a comment somewhere (iirc from CCP Yokai) regarding the Infiniband. It's basically on the backburner. I think the general issue was the (Infiniband) horse outrunning it's carriage (DB/data processes), so it doesn't really make sense to delve into it until things catch up to make it worthwhile.
Liandra Xi Amarr The New Era C0NVICTED	Posted - 2010.12.13 10:23:00 - [7] LXQ2 was shared with the Jita "node" supposedly during the epic fight that saw over 3000 in local and a few cpp devs podded trying to spy on us all lol.
Kalle Demos Amarr Hysteria Nexus	Posted - 2010.12.13 10:35:00 - [8] According to CCP this is all done during DT so im thinking when DT is removed completely nodes will auto reinforce themselves, no doubt you will whine after lag is gone anyway Originally by: Kool StoryBro <--- Originally by: CCP Spitfire Spam post removed. Random forum moments ftw
Myz Toyou APOCALYPSE LEGION	Posted - 2010.12.13 10:56:00 - [9] Oveur plugs some more of these into the cluster:
gn730qn4gjq3n4gqn348gnq3	Posted - 2010.12.13 11:07:00 - [10] Originally by: dkbjitawhore If you're lucky, yes. If your not lucky they simply make sure the designated system is on it's own server instead of having to share with other systems. If you're fighting IT in their own space then nothing gets done, especially if IT has to rep towers (which we all know works amazing in lag). Fixed.

Chesty McJubblies Gallente Center for Advanced Studies	Posted - 2010.12.13 12:40:00 - [11] Originally by: gn730qn4gjq3n4gqn348gnq3 Originally by: dkbjitawhore If you're lucky, yes. If your not lucky they simply make sure the designated system is on it's own server instead of having to share with other systems. If you're fighting IT in their own space then nothing gets done, especially if IT has to rep towers (which we all know works amazing in lag). Fixed. Your name has a typo in it. Suggestion: Remove the "new topic" button from everywhere apart from the list of topics section within a subforum. That'd save those with chronic hand/eye coordination some face.
RiotRick Samhain. viperfleet	Posted - 2010.12.13 12:44:00 - [12] They give the hamsters some red bull. --
Grimpak Gallente The Whitehound Corporation	Posted - 2010.12.13 12:59:00 - [13] Originally by: Sturmwolke Originally by: Akita T There was a lot of talk about "Infiniband"//HPC on the hardware side, which was supposed to make both of those things possible or at least somewhat easier or better (you still need massive software rewrites either way)... but I don't even know if that's a big focus at CCP anymore or not. If it is, it certainly hasn't received much public attention lately. There was a comment somewhere (iirc from CCP Yokai) regarding the Infiniband. It's basically on the backburner. I think the general issue was the (Infiniband) horse outrunning it's carriage (DB/data processes), so it doesn't really make sense to delve into it until things catch up to make it worthwhile. it would provide an interesting base framework to re-build the TQ cluster on, but yes. until you fix the software new hardware wouldn't make it go faster at all. --- Quote: The more I know about humans, the more I love animals. ain't that right.
Mr Dilkington	Posted - 2010.12.13 14:00:00 - [14] Originally by: RiotRick They give the hamsters some red bull. This but also they put a few more hamsters in the cage that powers your system as well.
Nicholas Barker Diabolus Ex Machina The Amazing Onjoi and his Educated Rodents	Posted - 2010.12.13 14:28:00 - [15] Originally by: Akita T Wrong. First off, no, Jita does NOT run on "several" nodes. You sure it doesn't have several for things like, the market, inspace, peoples hangers? I can't tell if i remember something being said like that, or if i just made it up in my head (i've got a hangover). ------ 0800-LAG-A-NODE

CCP Spitfire	Posted - 2010.12.13 15:39:00 - [16] Trolling post removed. Spitfire Community Representative CCP Hf, EVE Online


CCP Mindstar	Posted - 2010.12.13 16:06:00 - [17] Originally by: Wu Jiaqiu Do they place a Jita strength server to back up the system to boost its speed? Hey guys, chiming in here and hopefully can resolve a few myths The way we separate high load systems from the cluster (ie. Jita, mission hubs or predicted fleet fights) is to allocate a single system onto a node. This means that it gets the full resources available (ie 100% of a CPU Core) and that it will not cause lag in other solar systems. The concept of reinforcing a node, or having a "reinforced node" is a bit of a misnomer. It implies that we have some kind of awesome turbo button that we can push (that would be cool tbh). We call them dedicated nodes - ie. nodes with only one solar system on them. Node - A single instance of an EVE server process. Blade / Server - Physical server machine. Runs 4 Nodes in most cases. Most of the blades in the EVE cluster are running on the same spec machine (Dual CPU 3.3ghz Intel Xeon), but we have started trialling some later models of processor with a view to the future. One such blade is what got Jita to 2000+, and is currently in use for fleet fights. We have a lot of this stuff listed out in dev blogs as well, so if you want to read in more detail check out here, here and here. Oh yeah, and don't forget hamsters on red bull. --

Akita T Caldari Caldari Navy Volunteer Task Force	Posted - 2010.12.13 17:39:00 - [18] Originally by: Nicholas Barker Originally by: Akita T Wrong. First off, no, Jita does NOT run on "several" nodes. You sure it doesn't have several for things like, the market, inspace, peoples hangers? I can't tell if i remember something being said like that, or if i just made it up in my head (i've got a hangover). Well, you are _technically_ correct, some non-essential parts of it are executed on different machines nowadays. Then again, that's the situation with every other system in EVE anyway, to the point of not mattering, considering them "not quite the system itself". Hopefully almost everything will eventually be stripped away from the part running the physical simulation and placed on "character nodes". But that still doesn't solve the real problem. The bulk of what happens CPU-wise (namely, the physical simulation of space, which is what we actually do call "a system" when it really comes down to it), so, "a system" proper is still running on a single node // CPU core // execution thread, with no way (yet) to split THAT workload on two (or more) separate machines. _ Beginner's ISK making guide \| Manufacturer's helper \| All about reacting _
Ragnar256 Minmatar	Posted - 2010.12.13 18:20:00 - [19] So, if a predicted fleet fight gets the host system it's own node, what happens to the Blade/server when (due to ill fate) two large blobs collide without warning.
Akita T Caldari Caldari Navy Volunteer Task Force	Posted - 2010.12.13 18:27:00 - [20] Originally by: Ragnar256 So, if a predicted fleet fight gets the host system it's own node, what happens to the Blade/server when (due to ill fate) two large blobs collide without warning. Well, in not so many words, <deleted for disturbing imagery>. Basically, what happens to anything else when you try to stick in more than it can handle. _ Beginner's ISK making guide \| Manufacturer's helper \| All about reacting _

Yuki Kulotsuki	Posted - 2010.12.13 18:35:00 - [21] Originally by: Akita T Basically, what happens to anything else when you try to stick in more than it can handle. "Oh god! Oh god! Pull out! -fade to black-"? Originally by: CCP Lemur THIS IS GOD: ... IF YOU HAVE ANY MORE REQUESTS I'M AVAILABLE SUNDAY FROM 10:30 TO 12:00 TO RECEIVE YOUR PRAYERS.
Joe SMASH You Got A Purty Mouth	Posted - 2010.12.13 19:15:00 - [22] Originally by: Yuki Kulotsuki "Oh god! Oh god! Pull out! -fade to black-"? Funny enough, that is how my wife got pregnant. ----------------------------------- Originally by: Kali Zero Warp core stabilizers are like condoms. Nice and safe, but they make it a little less fun for everyone involved.

CCP Mindstar	Posted - 2010.12.14 11:09:00 - [23] Originally by: Ragnar256 So, if a predicted fleet fight gets the host system it's own node, what happens to the Blade/server when (due to ill fate) two large blobs collide without warning. This. Serious answer: If a fight happens in a system that is sharing with many others on a node (ie. if it is not predicted), it causes lag in all of the systems that are hosted on the node (not blade - nodes on the same blade rarely, if ever, affect each others performance). Additionally, because that node is hosting other systems as well, there are less resources available to the fleet fight itself, which usually results in it feeling slower. --

white kight Galaxy Punks Dead Terrorists	Posted - 2010.12.14 12:17:00 - [24] Originally by: CCP Mindstar Originally by: Ragnar256 So, if a predicted fleet fight gets the host system it's own node, what happens to the Blade/server when (due to ill fate) two large blobs collide without warning. This. Serious answer: If a fight happens in a system that is sharing with many others on a node (ie. if it is not predicted), it causes lag in all of the systems that are hosted on the node (not blade - nodes on the same blade rarely, if ever, affect each others performance). Additionally, because that node is hosting other systems as well, there are less resources available to the fleet fight itself, which usually results in it feeling slower. Is this why when you can be 20 jumps away from a massive fleet fight, and still have crippling lag? Originally by: CCP Greyscale :facepalm:
Jack Dant Minmatar The Gentlemen of Low Moral Fibre	Posted - 2010.12.14 12:29:00 - [25] Originally by: CCP Mindstar Node - A single instance of an EVE server process. Blade / Server - Physical server machine. Runs 4 Nodes in most cases. Most of the blades in the EVE cluster are running on the same spec machine (Dual CPU 3.3ghz Intel Xeon), but we have started trialling some later models of processor with a view to the future. One such blade is what got Jita to 2000+, and is currently in use for fleet fights. So, I take it, each core in the blade runs a single node process which contains a large number of systems. Why isn't it possible to have several node processes sharing a core, each of them with less systems? I presume moving a node across cores is much easier than moving a system across nodes, so you could do dynamic balancing.
Iamien Democracy of Klingon Brothers R.A.G.E	Posted - 2010.12.14 12:41:00 - [26] Edited by: Iamien on 14/12/2010 12:41:39 Originally by: Jack Dant Originally by: CCP Mindstar Node - A single instance of an EVE server process. Blade / Server - Physical server machine. Runs 4 Nodes in most cases. Most of the blades in the EVE cluster are running on the same spec machine (Dual CPU 3.3ghz Intel Xeon), but we have started trialling some later models of processor with a view to the future. One such blade is what got Jita to 2000+, and is currently in use for fleet fights. So, I take it, each core in the blade runs a single node process which contains a large number of systems. Why isn't it possible to have several node processes sharing a core, each of them with less systems? I presume moving a node across cores is much easier than moving a system across nodes, so you could do dynamic balancing. To have one node on multiple cores, they would have to multi-thread the node process, which to put it simply, would mean a major re-write for questionable returns(overhead of multi-threading and such.)
Jack Dant Minmatar The Gentlemen of Low Moral Fibre	Posted - 2010.12.14 12:51:00 - [27] Originally by: Iamien To have one node on multiple cores, they would have to multi-thread the node process, which to put it simply, would mean a major re-write for questionable returns(overhead of multi-threading and such.) You got it wrong. I meant several nodes per core, each of them with less systems. So instead of 1 node serving 40 systems on a single core, you have the same core running 40 nodes of a single system each. If a fleet fight breaks out, you move away the extra nodes to other, less loaded, cores (something the OS should do relatively cheaply), and leave the fleet fight alone in its own core. I guess there is a good reason for not doing this, just curious about why.
Iamien Democracy of Klingon Brothers R.A.G.E	Posted - 2010.12.14 13:10:00 - [28] Originally by: Jack Dant Originally by: Iamien To have one node on multiple cores, they would have to multi-thread the node process, which to put it simply, would mean a major re-write for questionable returns(overhead of multi-threading and such.) You got it wrong. I meant several nodes per core, each of them with less systems. So instead of 1 node serving 40 systems on a single core, you have the same core running 40 nodes of a single system each. If a fleet fight breaks out, you move away the extra nodes to other, less loaded, cores (something the OS should do relatively cheaply), and leave the fleet fight alone in its own core. I guess there is a good reason for not doing this, just curious about why. Multiple processes have more overhead than a single process. As far as dynamic load balancing, they are currently developing a way to re-map systems on the fly. If you participate in the mass tests on SiSi you will commonly get disconnected when they initiate the re-maps(The disconnect is just how they are doing it to start, one day they will try to make it seamless).
Manipulator General o.0	Posted - 2010.12.14 13:49:00 - [29]

Chribba Otherworld Enterprises Otherworld Empire	Posted - 2010.12.14 13:55:00 - [30] Imma space cowboy riding the nodes into infinity! Secure 3rd party service \| my in-game channel 'Holy Veldspar'




Pages: [1] 2 :: one page
First page \| Previous page \| Next page \| Last 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.