Speaking of surplus power I wonder if it would even be remotely technically possible to have players themselves contribute computing power to EVE during heavy loads. Obviously it would have to be calculations on things that wouldn't be considered a security risk for EVE yet are vital enough to warrant this.

If by some miraculous means this would be possible one could technically eliminate all kinds of lag cause more players in a fleet fight would equal more processing power. The new bottleneck would probably then be bandwidth instead considering the increased amount of data flying back and forth between Tranq and clients.

Meh...just musing on possible futuristic utopian solutions so don't mind me too much

Well this is why I created this thread. I don't want to make assumption when I can have facts and figures.

Due to recent events I think it is important to bring some focus to the hardware and software making New Eden what it is today.
I can remember in the past 3 years the servers were upgraded and EVE was down for a long time but it came back better than before. Maybe we need another upgrade or at least a better understanding of the daily problems CCP faces.


-The Questions-

• What are the specs of the current infrastructure the hosts EvE Online?


• Is there something better out there that could be bought or built?


• If something better dose not exist what prevents CCP from buying more of what they have currently and using that to supplement the load?


• What kind of software is used to facilitate the hosting of EVE? Can it be improved upon at a core level? Does something better exist?

If these questions could be answered I think it could cause a drop in the angry threads and misinformed posters while at the same time cultivating more informed and constructive conversations on the subject.

CSM candidate without access to a basic Google search?

What still worries me as a huge EVE fan looking to invest years in skilling toons, is that a published strategy for a scalable and dynamic resource allocating architecture & plan to get there does not appear to exist...

This troubles me...10 years in and there is still game affecting lag in the infrastructure & architecture at all?

"A CFC Hel died, apparently unable to turn his hardeners on due to TiDi. Well before this time, TiDi had hit the maximum, ten percent, and the lag was horrible; it took a carrier pilot nearly ten minutes to refit a smartbomb. Local was well over a thousand and climbing; it would peak at over two thousand and seven hundred."

and...

"As the CFC skyteam put it, lowsec has the worst hardware of the cluster; huge fleet fights should never be held there. Elise commented that the lag also affected his target calling as travel time for drones, such as fighter bombers, must be taken into account"

and...

"Vahilior JEFFRAIDER
Thanks for the video link, its brilliant. By the time I showed up all the titans were warped off and TIDI + lag made everything unplayable, great to see this perspective" http://themittani.com/news/asakai-aftermath-all-over-cobalt-moon

CCP, what is your strategy for a dynamic resource allocating architecture that will prevent this from happening in the future, especially as player base grows? How can you claim to be a single shard sandbox MMO when said sandbox has intolerable performance management issues, and like cat pooh hidden just under the surface of said sand, the more people that enter the sandbox causes more people to get pooh between their toes?

I think the max for current server solutions from IBM is 3.6Ghz, not counting the super-awesome-only-used-on-wallstreet kinda stuff.

The current ones on TQ are 3.3Ghz. I'd rather not have CCP upgrade the entire cluster for a mere 300Mhz increase.

The eight-core E7-2830 @ 2.13 GHz has three times the performance of a dual-core x5260 3.3GHz.

But the difference is less if the critical part of the software can't run in multicore/multithread.

You're saying the critical part of EVE's server-side software can't utilize multiple cores?

It's pretty well known (I thought) that CCP's server and client code is not multi-threaded.

It's pretty well known (I thought) that CCP's server and client code is not multi-threaded.

Just because I didn't see anyone post it, here is a video from last FanFest 2012 titled Growing EVEs infrastructure: http://www.youtube.com/watch?v=rj_hHIwInwk&list=SPDDA989F65CD6E98A&index=12

Wasn't referring to the client code. The server software can't utilize multiple cores?

Just because I didn't see anyone post it, here is a video from last FanFest 2012 titled Growing EVEs infrastructure: http://www.youtube.com/watch?v=rj_hHIwInwk&list=SPDDA989F65CD6E98A&index=12

While interesting as hardware geek pronz, it still doesn't speak to a strategic plan for a dynamic resource allocating architecture, or provide an update on how real the comment on coding underway for dynamic node allocation is progressing to fix lag issues.

The re-enforce node supercomputer that's dependent on players prior notification of an upcoming fleet battle may sound good from a CPU/RAM & 'only defense agencies use this IBM pwnzor server!' marketing hype, but in reality its kinda useless isn't it? Big fleet battles typically do not come with prior notice, they just happen...you know, that whole emergent single sandbox thing core to EVE?

The lack of confirmed & committed plan for a dynamic resource allocating architecture scares me, because TiDi and prior-notice-fleet-servers give the impression CCP is just tweaking at the edges.

And CSM, where are you guys on this? Gameplay mechanics are great and all, but if CCP can't get a scalable dynamic resource allocation architecture right in 10 years to prevent lag, don't we have bigger fish to fry than POS & SOV tweaks?