Why 2175 ?

Obviously your answer will be something like "experience shows that if we cap at 2175 then x won't happen", but I'm curious what the x is.

So theres no way the node can be overloaded permanently to like 4000 chars?

Being a bit thick here but what is the limit on one system?

So with the current size of battles that are happening in 0.0 where you have to implement Tidi to make it playable at all, and the fact that Jita is frequently being capped to save the node dying. Would you conclude that it is logical to assume that the current hardware that Eve is run on is insufficient for it's intended purpose going forward?

As been said before, the biggest limiting factor to eve right now is the old code that was designed for single core units, and is not very well suited to be scaleable. Back when eve was made the big things was more GHZ. Then a few years later, intel dropped the core two duo/dual core processor, so the trend switched from MORE GHZ to MORE CORES. If you have code, that's can't scale across cores very easily then it won't matter how many cores you have, it still won't efficiently use them.

I used this analogy before, TQ is like the smoothest, flattest, more frictonless and aero dynamic road you can get. The code is a pinto. You are trying to drive a pinto down this perfect raod and expect it to go like a farari. It can't, cause its a ******* pinto.

So they seem to be slowly fixing old code, like the crime watch, bounty and other fixes. With things like brain in the box in the works, and other fixes we should see a huge improvement. EvE REALLY needs a rewrite in areas that can not be changed over easily. Then would make a huge improvement. But there is still no other game out there that can event remotely handle 2k+ people in one system doing things. at one time jita had I think 5k in it and was running decently, but that was ages ago and I could be wrong.

Does a cpu compress data that's getting shared in between nodes, to reduce bandwith? I'm asking if you're trading cpu cycles for bandwith. I imagine that even if the eve server software doesn't run multithreaded, there are ways to relay the data to different cores and having them compress them before sending.

I'm just asking. I realize that you have tons of gigabits travelling around and so I figure that even one byte less in every data package is a win already.

Why not run a cap based on the TiDi factor instead or a fixed number of pilots ?

One thing someone at CCP has mentioned before is how they have to change the population cap when expansions come out, because the percentage of docked players in Jita changes for a brief period after an expansion comes out. A TiDi based lock on the gate would automatically adjust to that and to any server side improvements.

Stop and make the calculations. The ammount of data that needs to transit OUT and Into jita is rather small in fact . PRobably the bottleneck profile in jita is way different than ina fleet battle where the massive syncronized jump ins are the main problem.

The traffic at jita gates while substantial does not cause so many session changes in the same server tick as we see when 800 ships cyno into a system (based solely on my observations). Therefore I must conclude that the dock/ undock party at Jita 4-4 is so detrimental to performance?

All this I say thinking that surely market transactions are not bound to the same issues of "single cpu" bottloneck per system since the market gives impression of being processed at region level.

No. In other words, the spammers don't cause that much sever load so they are not the reason legit players can't get in. If they were removed, the cap would just be lowered to maintain the same node load and you'd still not get in.

GǪa cap that is set to accommodate the 1175 non-spammers and the 1000 spammers. If you were to remove the spammers, somehow, the cap would be set to 1175 instead since that's how much load the system can take (after all, the spammers don't contribute to that load). The cap is set based on the load generated by the players logging or jumping into the system. Removing the people that add nothing to that load would just reveal that the actual cap is 1175, so you'd still have just as many slots available for the non-spammers and you'd still not get in.

Your assumption isn't based on any facts. You just think that's how it works. Unless Explorer comes in here to confirm that the cap is there with the fact that x % of it will be spammers; you're talking rubbish. It makes just as much sense to say that spammers removed from the system free up space for legitimate players.

You know what's kind of irritating?

Jita is capped at 2175 only.

All market transactions are done on a different node.
All chatting is done on a different node.

This actually removes two of the most engaged things happening in Jita 4-4, right?

... and that's kind of irritating. I'd bet that most people believe that trading and chatting are the things
actually slowing the node down the most.

So ... what's actually slowing down the node so much?

The session changes ?? Jita does have quite a few gates and there are always several people
at the New Caldari and Perimeter gates. Then we have several stations too.

Is it true?

Why did I write 'only' in my second line? Because 2175 isn't that much, if you put it into context.

If you look at the bloodbath of B-R5RB and think of all the things that are going on, including amount of ships, drones, session changes due to explosions/jump-ins, people clicking modules, PHYSICS!, etc etc ...

Yes, there's TiDi at 10% ... but the node wasn't even reinforced !

TL;DR: There was SO much more going on in B-R5RB which ran on a non-reinforced node with population count comparable to the one of Jita for quite a while. Jita runs on a reinforced node AND has two of the most happening things in system outsourced onto a different node!

What's actually going on that you need to have a cap as low as 2175?

Lots of people logging in, jumping in, and undocking, which requires large amounts of calculations to set up their in-space object. Unfortunately, it's also not something that TiDi is a very good countermeasure for, which is why everyone is (or was) hoping for some progress on the brain-in-a-box project that would sideload all those calculations onto a separate server.

Fleets generate a completely different kind of load. Sure, when it jumps in, all those calculations have to be run, which creates a momentary spike, but once that's done, everything else is just the standard ongoing simulation running at whatever speed it needs to run at. That load is generally much smaller per ship than the complex task of fetching all character, ship, module, etc. data, mashing it together and producing a single object for the simulation to handle.

Why should NPCs be handed a job players can do just fine on their own?

Not exactly true the blade with the core that jita runs on has 3 other cores that can run at the same level as jita. one is used for systems that a battle node as been requested for and the other two are assigned other trade systems or other high usage systems.

You do realise that whether I go to Jita or I get another player to go to Jita (because that's where they're going to go for the cheapest price) that's still the same number of people going to Jita.

Just reposting this.

This is such a load of garbage. Is Tippia your forum alt or something, you tend to agree with her nonsensical statements quite a lot.

Why do I say that. Because quite often Jita is full and is functioning quite well but the queue is still active. You cannot base a performance cap on number of pilots in system when the number of pilots in system != server load since the number of people undocking docking and jumping is variable.

Theoretically you could have 2100 people in Jita all docked and AFK and queues on the gates when the server is not being stressed at all. There is no pretend dynamic actual cap only a static 2100 cap based on pilots in Jita.

Unless we moved to another universe where logic no longer works they're both wrong. As I stated if 2100 people are AFK then if Tippia is correct then the actual cap must be 0. That's illogical.

If you have a static cap you can not theoretically have a dynamic cap. Its like trying say black is white. Dev or not its illogical to try to say that.

If his small percentage of people are spammers is even 50 people that's still 50 people waiting at a gate. If there's only 50 people afk that's 100 people waiting at a gate. Since the amount of spammers and afk'rs is likely higher and likely varies over peak times then its even more incorrect.

Correct me if I'm wrong, but what you're effectively saying is,when it comes to Jita cap, one person doesn't necessarily mean one person. I've always been curious about it.

You guys basically break it down to a measurement (lets call them lag bubbles). One person's actions may mean they're worth 5 "lag bubbles", while another person may only be worth one.

Jita has an average idea, overall, of how much lag each player is worth. That translates into the cap. You get traffic controlled, when Jita is unexpectedly handling more "lag bubbles" than it should. Lets say, in example, one person on average equals two lag bubbles, but the last guy jumping in, he's generating 5 bubbles. Jita has to block gates, until those excess units bleed off somehow.

I have one question about that. When it's deployed, can we all get an inaugural Brain in a Box collectible item?

It would go great with corpse collections.

Why do I say that. Because quite often Jita is full and is functioning quite well but the queue is still active. You cannot base a performance cap on number of pilots in system when the number of pilots in system != server load since the number of people undocking docking and jumping is variable.

Theoretically you could have 2100 people in Jita all docked and AFK and queues on the gates when the server is not being stressed at all. There is no pretend dynamic actual cap only a static 2100 cap based on pilots in Jita.

Well do you know what the 100 spammers are doing besides just spamming? What is to say they are doing less than an 100 active pilots that just want to dock to buy/sell and check stuff or just chat with the others?
To simple just dismiss leads me to believe you are happy to have spammers in the game.
How much processing is on everyone's block lists when spammers are blocked?
You get rid of spammers, then you might get new players and probably more people buying and selling stuff (you know a better economy).
Mind you with the amount of spammers, CCP could lose a lot of revenue lol
No matter what - make sure that block list is capable of holding 1000+ since from what I have witnessed - every spammer toon has at least 10 alts - lol

Any plans to destroy Jita - make the surrounding systems low sec (like Yulai) and creating a general panic as a new market hub is created. This would cause so much excite and market craziness. Please do.

Also what is the difference betwen a fleet fight node and a regular node?