These forums have been archived and are now read-only.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.
12Next page
 

CCP pay attention!! Your hamsters can be saved...at least a few anyway
Author
Previous Topic Next Topic
Natassia Krasnoo
R3D SHIFT
#1 - 2014-03-11 15:03:40 UTC
So I was reading the news this morning when I ran across this in the tech section.

This definitely sounds like something CCP could take advantage of. More throughput and data bandwidth to the servers could never hurt. While it's not the ultimate solution to the single threaded backend of EVE, it could probably help out the hamsters quite a bit and possibly spare a couple of them from dying of exhaustion.

Just a thought, but think of the hamsters CCP, think of the hamsters.
Batelle
Federal Navy Academy
#2 - 2014-03-11 15:11:04 UTC
hey, a suggestion about saving hamsters that for once does not require rewriting the entire codebase. \o/

"**CCP is changing policy, and has asked that we discontinue the bonus credit program after November 7th. So until then, enjoy a super-bonus of 1B Blink Credit for each 60-day GTC you buy!"**

Never forget.
Aerozzz
Afragoi Ltd
#3 - 2014-03-11 15:31:42 UTC
Natassia Krasnoo wrote:
So I was reading the news this morning when I ran across this in the tech section.

This definitely sounds like something CCP could take advantage of. More throughput and data bandwidth to the servers could never hurt. While it's not the ultimate solution to the single threaded backend of EVE, it could probably help out the hamsters quite a bit


No it won't.

/thread
Pak Narhoo
Splinter Foundation
#4 - 2014-03-11 15:33:01 UTC
Aerozzz wrote:
Natassia Krasnoo wrote:
So I was reading the news this morning when I ran across this in the tech section.

This definitely sounds like something CCP could take advantage of. More throughput and data bandwidth to the servers could never hurt. While it's not the ultimate solution to the single threaded backend of EVE, it could probably help out the hamsters quite a bit


No it won't.

/thread




Such an elaborate reply.
Do tell, why no?
Victor Andall
#5 - 2014-03-11 15:37:39 UTC
Data Transfer != Processing Power

Elaborate enough?

I just undocked for the first time and someone challenged me to a duel. Wat do?

19.08.2014 - Dinsdale gets slammed by CCP Falcon. Never forget.
Lucas Kell
Solitude Trading
S.N.O.T.
#6 - 2014-03-11 16:19:57 UTC
Pak Narhoo wrote:
Aerozzz wrote:
Natassia Krasnoo wrote:
So I was reading the news this morning when I ran across this in the tech section.

This definitely sounds like something CCP could take advantage of. More throughput and data bandwidth to the servers could never hurt. While it's not the ultimate solution to the single threaded backend of EVE, it could probably help out the hamsters quite a bit
No it won't.

/thread
Such an elaborate reply.
Do tell, why no?
Their bottleneck is not data throughput. If it were a bottleneck, it would be unlikely to be their internal network, and unless they plan on replacing all the cabling and hardware from the CCP servers up to my PC, it really wouldn't matter.

To be honest the whole thing is pretty useless unless you are bulk copying terabytes of data. This is really just another of those "look how high our numbers go!" kinda things. The whole reason CCP are in the situation they are is because when everyone was saying "look how high our CPU speeds can go!" they invested heavily in it. Then that peaked and the trend turned to multithreading, allowing process to perform more than one task at a time and CCP are stuck where they are.

The Indecisive Noob - EVE fan blog.

Wholesale Trading - The new bulk trading mailing list.
Rainbow Dash
The Scope
Gallente Federation
#7 - 2014-03-11 16:53:01 UTC
/thread
Xercodo
Cruor Angelicus
#8 - 2014-03-11 16:54:48 UTC
Btw I was made aware of how hadoop works recently and was wondering if there was an IT guy that could explain why this wouldn't work for TQ because it sounds like a perfect system regarding scalability and I would have thought CCP have already been made aware and looked into it.

Is it simply because they have the own server node management system so tightly intertwined into the code that they simply cant let hadoop handle it without major rewrites?

The Drake is a Lie
Unsuccessful At Everything
The Troll Bridge
#9 - 2014-03-11 16:57:45 UTC
CCP, your hamsters can be saved...all you need is all new hardware with the correct ports in order to utilize this miracle technology. This sounds like a simple and cheap investment/upgrade and should be implemented immeadiately.




Since the cessation of their usefulness is imminent, may I appropriate your belongings?
Serene Repose
#10 - 2014-03-11 17:13:47 UTC
I'm a big fan of cheap solutions to imaginary problems!

We must accommodate the idiocracy.
Aerozzz
Afragoi Ltd
#11 - 2014-03-11 17:15:07 UTC
Xercodo wrote:
Btw I was made aware of how hadoop works recently and was wondering if there was an IT guy that could explain why this wouldn't work for TQ because it sounds like a perfect system regarding scalability and I would have thought CCP have already been made aware and looked into it.


I'm not sure you understood what Hadoop is / does. It's pretty much as useless for EvE (as far as the current major EvE performance issues are concerned) as the new 800Gbps Intel interconnects.

PS: Just noticed, "800Gbps per second" in the pcworld article. /facepalm
Bagrat Skalski
Koinuun Kotei
#12 - 2014-03-11 17:16:24 UTC
Free the hamsters! FREEEDOOOM!

Custom ship skins, character creator style "repaint". Bring it!
Xercodo
Cruor Angelicus
#13 - 2014-03-11 17:38:49 UTC
Aerozzz wrote:
Xercodo wrote:
Btw I was made aware of how hadoop works recently and was wondering if there was an IT guy that could explain why this wouldn't work for TQ because it sounds like a perfect system regarding scalability and I would have thought CCP have already been made aware and looked into it.


I'm not sure you understood what Hadoop is / does. It's pretty much as useless for EvE (as far as the current major EvE performance issues are concerned) as the new 800Gbps Intel interconnects.

PS: Just noticed, "800Gbps per second" in the pcworld article. /facepalm


See I figured that answer was inevitable, but what I wanna know is why.

Hadoop looks like a great system for evenly distributing the load of the whole server over all of the nodes instead of CCP needing to reinforce specific nodes like Jita.

Is it because the node processing system is so deeply ingrained into the architecture of the game code already?

The Drake is a Lie
Dersen Lowery
The Scope
#14 - 2014-03-11 17:49:04 UTC
Xercodo wrote:
Is it because the node processing system is so deeply ingrained into the architecture of the game code already?


It's because the game engine is still largely single-threaded per system, and thus a terrible candidate for any kind of generic distributed-computing model. Even if the various efforts to break it up into separate threads succeed, it's still not going to result in a model that can be broken up into nice little pieces and distributed widely: one of the major bottlenecks is the bit of code that assembles all the individual calculations into a coherent model of what's happening on each grid in a system, and that's not the kind of work that factors well.

Proud founder and member of the Belligerent Desirables.

I voted in CSM X!
Lucas Kell
Solitude Trading
S.N.O.T.
#15 - 2014-03-11 17:59:05 UTC
Xercodo wrote:
See I figured that answer was inevitable, but what I wanna know is why.

Hadoop looks like a great system for evenly distributing the load of the whole server over all of the nodes instead of CCP needing to reinforce specific nodes like Jita.

Is it because the node processing system is so deeply ingrained into the architecture of the game code already?
CCP do use Hadoop. They use it to process server log files and other related data processes.

The thing with Hadoop for the rest of the game though is that the rest of the game isn't designed to work in a distributed environment. If it were, it would be considerably faster (performance wise) for them to design their own distributed processing framework rather than relying on Hadoop. To be honest, they wouldn't really need to distribute processes across multiple servers. If they could utilise multiple cores on a single machine they could leverage considerably more processing power per node.

The Indecisive Noob - EVE fan blog.

Wholesale Trading - The new bulk trading mailing list.
Real Serious PVPer
Doomheim
#16 - 2014-03-11 18:06:20 UTC
In space, there are no hamsters- Just PVP

[b] Serious about being serious- Putting the "P "into PVP one fight at a time.  MUFC[/b]
Pix Severus
Empty You
#17 - 2014-03-11 19:01:35 UTC
Here's an article that will give you some more information on the server.

MTU Hunter: Latest Entry - June 12 2017 - Vocal Local 5

MTU Hunting 101: Comprehensive Guide
Xercodo
Cruor Angelicus
#18 - 2014-03-11 20:51:04 UTC
Lucas Kell wrote:
Xercodo wrote:
See I figured that answer was inevitable, but what I wanna know is why.

Hadoop looks like a great system for evenly distributing the load of the whole server over all of the nodes instead of CCP needing to reinforce specific nodes like Jita.

Is it because the node processing system is so deeply ingrained into the architecture of the game code already?
CCP do use Hadoop. They use it to process server log files and other related data processes.

The thing with Hadoop for the rest of the game though is that the rest of the game isn't designed to work in a distributed environment. If it were, it would be considerably faster (performance wise) for them to design their own distributed processing framework rather than relying on Hadoop. To be honest, they wouldn't really need to distribute processes across multiple servers. If they could utilise multiple cores on a single machine they could leverage considerably more processing power per node.


Thank you this was the sort of answer I was looking for.

And it turned out to be a kind of hybrid of what I was originally thinking; that the code is structured in such a way that hadoop doesnt offer much benefit.

I'd think if CCP were able to break up the processing system on a per player basis instead of a per system basis you could distribute the load evenly over all the nodes. I just don't have any idea what sort of implications his has for communication between nodes though as a fleet fight happens involving half a dozen different nodes...

Either way I eagerly await the release of brain in a box :D

The Drake is a Lie
TigerXtrm
KarmaFleet
Goonswarm Federation
#19 - 2014-03-11 21:02:28 UTC
Natassia Krasnoo wrote:
So I was reading the news this morning when I ran across this in the tech section.

This definitely sounds like something CCP could take advantage of. More throughput and data bandwidth to the servers could never hurt. While it's not the ultimate solution to the single threaded backend of EVE, it could probably help out the hamsters quite a bit and possibly spare a couple of them from dying of exhaustion.

Just a thought, but think of the hamsters CCP, think of the hamsters.



Bandwidth isn't the bottleneck that strains the hamsters. It's processing power. Do your homework.

My YouTube Channel - EVE Tutorials & other game related things!

My Website - Blogs, Livestreams & Forums
Joshua Foiritain
Coreli Corporation
Pandemic Legion
#20 - 2014-03-11 22:04:03 UTC
Xercodo wrote:
I'd think if CCP were able to break up the processing system on a per player basis instead of a per system basis you could distribute the load evenly over all the nodes. I just don't have any idea what sort of implications his has for communication between nodes though as a fleet fight happens involving half a dozen different nodes...[

I dont think you understand how programming works P

The Coreli Corporation is recruiting.
12Next page