These forums have been archived and are now read-only.

The new forums are live and can be found at https://forums.eveonline.com/

EVE General Discussion

 
  • Topic is locked indefinitely.
 

Time Dilation feedback

First post First post
Author
GeeShizzle MacCloud
#101 - 2012-01-22 19:18:19 UTC
Zleon Leigh wrote:
Jarnis McPieksu wrote:
Zleon Leigh wrote:
CCP Lemur wrote:
Zleon Leigh wrote:
So - we're seeing TiDi engage in empty systems and in Jita. Fail? Oh wait!! By engaging it in empty systems you can lower how much hardware you need to field. Guess returning nodes to vendor is next step after layoffs....


I'm not a Veritas but I assume that since some nodes host a bunch of low load systems at once TiDi would kick in for all of these when one gets a sudden spike of activity. Or would you rather have the ones just screwed without any indicators? And I'm sure we are not giving any nodes back rather getting better ones like we have always been doing in the past and will continue to be doing in the future.



I'm in a 0.0 system with 17 pilots in it, several docked in station, NO FIGHTING, peaceful as can be - TiDI kicked in. LAME.




I think you fail to understand that single physical server can run multiple solar systems.

In the case of this huge fight, reportedly the same node (server computer) ran 75 solar systems. One of them has a huge fight and all 75 get time dilated because they are all on the same server.


LOL Trust me... I understand. There is no reason for TiDi to kick in for all systems on a node - that would just be stupid. You dilate the stressed systems and fork off the low end systems to other nodes. You don't turn the clock down on the entire CPU stack.





its VERY evident that you dont understand! ill put it as plainly as i can...

a metric **** ton of systems are handled by 1 server node and all of eve are separated into these big clusters of systems that are handled by one node or another. and by that i dont mean all of eve is handled by 2 nodes, theyre handled by a fair few nodes.

A server node is a term for server blade, the ones u see in rack cases in big companies, its like your computer at home but more likely, quite a bit more powerful. Almost everything a player does in game in one of these clusters of systems are handled by this one computer. please PLEASE note the use of the word almost in that sentence. Certain things like requiring info from databases (eg market transactions etc...) can request info from other nodes.

but the main point is that theres 1 computer with only a finite amount of processing power able to process what happens in those systems.

before Time DIlation when theres a fleet fight occuring on one of those systems its like back at home on your computer, 600 or more people at a house party wanting to open and use 600 different applications all at the same time expecting it to work and load everything straight away and run smoothly. some will work before others and some people only wanting to write something on notepad will take longer to do so than sum1 opening up a file on photoshop.

You could look at the problem by saying we'll process stuff one at a time and people will just have to wait. this process of waiting for your turn is called SERVER LAG. people hate it cause it means if your fighting in a time sensitive battle, you can die way before ever realising it.

Time Dialation is basically a process that says "woah, calm the **** down, i can process all of your stuff at the same time, just bit by bit. its not ideal i know, and things will take longer to do, but you'll all see stuff happening sooner rather than later."


so just to clarify...

TiDI can only kick in on all systems on the node cause theyre all handled by the same computer resources aka 1 physical CPU + RAM etc...

Eve Onlines Server Architecture cant remap players to different server blades without forcing them to disconnect. and if uve got like 75 systems on that server blade, thats quite a lot of systems to push to other places. and possibly a fair few people that WILL complain that theyre being randomly disconected from eve for no apparent reason. that kinda thing can cause people to loose expensive ships, for example if theyre in a DED complex or an Incursion.

Also theres no stack of CPUs on one server blade. the server architecture is not multi-cpu compatible, and the servers clock speed doesnt go down when TIDi is active.. if anything its running at absolute max clockspeed possible.


... damn i think ive just been trolled :/
Zleon Leigh
#102 - 2012-01-22 19:22:08 UTC
Apparently so... I used to do CPU hardware design for supercomputers... :)

Incarna - Newest business example of mismanaged capital. CCP - Continuing to gank independent PI producers every day

PvP's latest  incentive program ** Unified Inventory **  'Cause you gotta kill something after trying to use it

John Maynard Keynes
Science and Trade Institute
Caldari State
#103 - 2012-01-22 19:38:06 UTC
On the one hand TiDi is a great feature, however it introduces new problems the people will be bitching about sooner or later.

The time on the TiDi node runs slower, so that it becomes easier to send reinforcements in and continue the fight infinitely. (1min on TiDi node = 10 minutes on other nodes).

It is now easier to react to certain situations. You can see dictors coming closer switch your targets in time so that they can't bubble your supers. You have more time to broadcast for reps as soon as you see that you get yellowboxed.

This happens when you try to fight the symptoms and not the cause.

New Eden became "too small" due to jump drives and bridges. Even if you rf 100 structures all around eden, the defender can wait till the enemy arrives at one of them and deploy his entire blob to this particular system. This would be by far more difficult without the awesome mobility we have now.
Tallian Saotome
Nuclear Arms Exchange Inc.
#104 - 2012-01-22 19:42:10 UTC
Zleon Leigh wrote:
Apparently so... I used to do CPU hardware design for supercomputers... :)

Then you should realize that distributed load processing across an HPC cluster using a single threaded application(which the eve server is, being written in stackless python) would require a complete rewrite of the core system, mean many man years dedicated to just that task, and then more dedicated to re-writing everything else to work with it.

Its actually a current work in progress, as they convert code to C++(last I heard) and that will resolve a large number of the core issues like this. Til then they have to maintain the current codebase actively, and continue work on it to keep the game alive and kicking.

But, I don't have to tell someone who used to design supercomputers that Roll

Inappropriate signature removed, CCP Phantom.

Zleon Leigh
#105 - 2012-01-22 19:47:20 UTC
Correct. So what the should have been done is limit fleet engagements to what was practical for the given software arch and hardware that meets business goals (can't afford super-shiny boxes) and satisfies the customer base (not just one particular group of players). Instead they continue to reach for the pie....

Incarna - Newest business example of mismanaged capital. CCP - Continuing to gank independent PI producers every day

PvP's latest  incentive program ** Unified Inventory **  'Cause you gotta kill something after trying to use it

Tallian Saotome
Nuclear Arms Exchange Inc.
#106 - 2012-01-22 19:50:42 UTC
Zleon Leigh wrote:
Correct. So what the should have been done is limit fleet engagements to what was practical for the given software arch and hardware that meets business goals (can't afford super-shiny boxes) and satisfies the customer base (not just one particular group of players). Instead they continue to reach for the pie....

And how do you limit engagements? hard cap on number of people in a system? We will just pile that many people into the system so no one can challenge us.

Limiting engagements has been batted around for years, and no one has come up with a solution that can't be gamed.

Inappropriate signature removed, CCP Phantom.

Zleon Leigh
#107 - 2012-01-22 19:55:19 UTC
Dunno.. but no matter what solution they reach for this way it can have the same consequence - ever larger groups will break it, with ever more catastrophic results.

Incarna - Newest business example of mismanaged capital. CCP - Continuing to gank independent PI producers every day

PvP's latest  incentive program ** Unified Inventory **  'Cause you gotta kill something after trying to use it

Tallian Saotome
Nuclear Arms Exchange Inc.
#108 - 2012-01-22 20:02:35 UTC
Zleon Leigh wrote:
Dunno.. but no matter what solution they reach for this way it can have the same consequence - ever larger groups will break it, with ever more catastrophic results.

TD, in theory, fixes that(or at least helps mitigate it).

With the current technical limitations(put in place by people who mostly don't work there as devs anymore) they can't move systems from one node to another without kicking everyone in the affected systems off the game. We are in a be patient holding pattern for more significant fixes(as I said, you can see from dev blogs they are converting the codebase to C++ slowly but surely, which will allow them alot more options) and the bandaid fixes and mitigation fixes will keep coming to try and keep up with how we break the system.

Oh yeah, you also have to remember that CCP is 100% a Microsoft shop, so all the servers are running on a windows server platform, the database is MSSQL, etc etc etc. In that building, its either microsoft, or in house, which might explain some of the other behavior we see.

Inappropriate signature removed, CCP Phantom.

GeeShizzle MacCloud
#109 - 2012-01-22 20:54:25 UTC
and im pretty sure that the mssql thing is an 8 year old issue too... cause at that time for a start up company it probably seemed a sensible choice given their location, assets and available employee skillpool.

im fairly certain every dev there knows now that unix/linux would probably be greatly preferable, but eves server codebase i guarantee u cant be ported to it without a significant re-write! u might find that Dust514 will be written server side to be used on unix/linux and just coded to allow information to be translated between eve and dust servers, and most likely only possible due to outside investors (eg Sony).

but until ccp have the funds available to dedicate significant dev man years to Eve Online itself, fixes to the current system are the only viable option on the table. and TiDi is a shining example of what can be done on a limited system with some creative thinking and coding.

i for one think Veritas is a f**king legend =)
David Laurentson
Laurentson INC
#110 - 2012-01-22 21:59:06 UTC
Zleon Leigh wrote:
Dunno.. but no matter what solution they reach for this way it can have the same consequence - ever larger groups will break it, with ever more catastrophic results.


Catastrophic like "some people elsewhere experience lag"?

That's not very catastrophic at all. I'm pretty much certain that previously a 1k man fight on a node split 75 ways would have crashed the entire node a year ago. Now, it slows down a bit.

Oh, as they say, noes. Given the choice between 0.1% of EVEs star systems slowing down for a while, and never having any four-figure-pilot fleet fights, I think this'll do.
Finde learth
Republic Military School
Minmatar Republic
#111 - 2012-01-22 23:46:40 UTC
DeLaBu wrote:
It would be nice to see a video of time dilation in action in a real fight (not one of the SISI tests).

Anyone has a link to something like that?

http://www.youtube.com/watch?v=4L4-gSIRssw&feature=player_embedded
Jita Rich
Alternative Life Incorporated
#112 - 2012-01-23 00:47:01 UTC
As a PVP'r:
TiDi is freaking awesome. Gate movement seems smoother. Modules actually work in fights (~400 total pilots). It does seem to come up pretty often though. Especially when bridging through titans.

As a PVP'r carebearing:
Complete crap. It's really hard to earn isk under time dilation. Unfortunately TiDi is limited to within the eve server and does not apply to real life. Therefore, my time is being devalued in the game.


I think TiDi could scale better if it worked on a per system basis rather than per node. I understand this means less performance, but it just doesn't make sense to be slowed down if you're 10 jumps away from a 100man moving fleet. An alliance with many players could easily use the current implementation to meta by slowing down other players. They wouldn't even need more than rookie ships. Basically its DDoS w/ spaceships.
GeeShizzle MacCloud
#113 - 2012-01-23 01:31:45 UTC
Jita Rich wrote:


I think TiDi could scale better if it worked on a per system basis rather than per node. I understand this means less performance, but it just doesn't make sense to be slowed down if you're 10 jumps away from a 100man moving fleet. An alliance with many players could easily use the current implementation to meta by slowing down other players. They wouldn't even need more than rookie ships. Basically its DDoS w/ spaceships.


well it can work on this basis, but it requires ccp to remap a single system to the reinforced node at a downtime before the fight takes place. unfortunately that means people predicting a fight to take place when theres no actual guarantee it will.

even if the fight does take place on the reinforced node as predicted, generally theres several fights that take place in and around the area that can collectively impact the rest of the mapped systems.

so you see it is possible, its just not do-able in a dynamic ad-hoc way currently.
Nova Fox
Novafox Shipyards
#114 - 2012-01-23 01:46:28 UTC
Wonders if the next hurdle is going

Dynamic Node Remaps

DynR

So that when massive amounts of lag pops up on one node they start shuffeling off the low load solar systems off of it. (much better than shoving a high load to a dedicated first.

Dust 514's CPM 1 Iron Wolf Saber Eve mail me about Dust 514 issues.

Salpun
Global Telstar Federation Offices
#115 - 2012-01-23 02:01:17 UTC
Nova Fox wrote:
Wonders if the next hurdle is going

Dynamic Node Remaps

DynR

So that when massive amounts of lag pops up on one node they start shuffeling off the low load solar systems off of it. (much better than shoving a high load to a dedicated first.

Depends how many large fleet fights happen in the next month. I think the issue will be solved faster with core game play decisions/ functionality. All this hinges on when the crime watch rewrite gets done. Take that off the main core, fix the client lag when you have a large number of pilots in the overview, possibly while breaking it up into more watch list type windows and then we can go back to server side issues.

That sayed the earlier DynR gets in the game the better.

If i dont know something about EVE. I check https://wiki.eveonline.com/en/wiki/ISK_The_Guide

See you around the universe.

Aineko Macx
#116 - 2012-01-23 07:59:02 UTC
Tallian Saotome wrote:
Its actually a current work in progress, as they convert code to C++(last I heard) and that will resolve a large number of the core issues like this.

Most execution time is already spent on C++ code held together by Stackless Python "glue". The problem here is Pythons GIL, that effectively serializes execution. Here is a devblog with some snippets.
DarkAegix
Acetech Systems
#117 - 2012-01-23 12:39:48 UTC
If someone actually filled the fleet fight notification form to gain their own special super node for that massive fleet fight, there would be a lot less baseless crying in this thread.

Also, time dilation across an entire cluster is better than everyone in the cluster having their modules break and client crash.
Vegare
Bitslix
Lolsec Fockel
#118 - 2012-01-23 13:18:48 UTC
I participated in two ~1400 man fights last weekend and have to say: TiDi is awesome!

Everyone who was there knows fights like those would have not been possible not long ago.

To all those whiners: obviously there are lots of gameplay issues to being dilated down to 0.1 but its IS gameplay contrary to being blackscreened.
Kyt Thrace
Lightspeed Enterprises
Goonswarm Federation
#119 - 2012-01-23 17:07:05 UTC
I too was in Branch this past weekend in ~1400 man fleet fights.

TiDi is the bomb, it rocks, one of the best additions for big blob fests.

I actually got on my first titan kill which was because of TiDi & the fact the titan just couldn't log off like in the old days.

I can only see TiDi getting better & better in the future. Way to go CCP.

R.I.P. Vile Rat

Lyron-Baktos
The Scope
Gallente Federation
#120 - 2012-01-23 18:00:26 UTC
experienced TiDi over the weekend and was in a system, maybe more than 1, and it was just out gang of 9 people. It sucked