Commander: Stay logged in! Whatever you do, cap the system to keep enemy out!CCP has since long advertised the fact that they have all of their players playing on the same server. While this is close to the truth it is not the entire truth. Lets investigate this further. Infact CCP has a large number of servers all interconnected which runs the EVE universe. A so called clusterer. Its infrastructure built so that players and systems are loadbalanced over a large number of fast servers and always striving to maintain an even balance to handle playeractivity. Some systems are less prioritized and these systems can get to share one server. It's when a sudden change such as warfare or special events occur that sudden lag occurs in these systems which can all be affected when one of the servers handling many systems get hampered with workload for one particular server. So, what is known: 1 system cannot be run over several servers, but many systems can be run on one server. This kind of architecture is limited by the coding practises implemented by CCP from start which in turn relies on the hardware to give a smooth experience for us players. CCP dedicated countless hours on improving the coding and squashing bugs which may decrease performance. The other measures CCP has taken is to upgrade the actual hardware to sustain more and more players in one system. Perhaps CCP never envisioned a thousand players in one system, but counted on them being spread out over many systems which they could split up on multiple servers but at the moment it seems like CCP has hit a wall. They simply cannot improve the code fast enough to get more players into one system, and the current hardware available only manages to handle as much as a couple of hundred in the same system. We've seen the word "reinforced" used by players and CCP when dealing with lag. It has been CCP's methods for years to handle systems with high workload. Simply to dedicate one server in the cluster to that particular system. But now it seems that this wall that CCP hit can't simply be helped by: reinforcing the system on a dedicated server, improving code and getting new hardware. Their final option to avoid cluster crashes is simply to put virtual limits on the system as we've seen in the recent war. This is not an ideal situation for either of the sides since it will have many implications which I will come to soon. CCP has to solve this either by finding a way to improve the architecture of their game. Making each system more decentralized or to rely on super-computers readily available within the nearest future. Or perhaps some option which I've yet to see CCP come up with. Another aspect of this problem is the network-traffic outbound from each system. All the traffic sent to players will cause lag since each calculation, weaponfire, shipchanges etc will have to be propagated to each client. Multiply this with a couple of hundred players and you will see that players will get issues clientside with networktraffic (limited by their own network access & computers for calculating _everything_). This virtual limit CCP chose has been to limit the system to a max-amount of players. This limit has been put in place a bit randomly previously but seems to become the norm now to avoid crashing the node entirely. Perhaps this will save one nights sleep for a sysadmin on call but in the end it will cause a bigger headache for CCP if it turns out as I think. The following will happen: - Warfare tactics will change. The people with the most players in system during a long period of time will win. If you manage to lockout your enemy with the help of the virtual limit the battle is won even before any turrets are fired. Simply smack up a POS just as you put the enemies POS into reinforced. Have it outfit as a deathstar with lots of strontium with just enough time to survive until you can remove the enmies POS. During this whole time you strive to maintain the largest amount of players in system. Doesnt matter how much forces you got outside the systems. Just as long as you can cap the system you're trying to defend/attack. Just imagine the login-queues and "Login battles" during these days! Both sides will have their forums spammed with "Make sure you keep your pilots online even if you're afk! Fit a cloak, go to our pos. Just be online to keep enemy out!!!" . Result: BIG headache for CCP when either of sides complain about this new tactic. More serverload, login-queues and all that crap associated with high playernumbers. CCP must realise that these virtual limits are the worst kind of tactic they can chose. They must remove these virtual-limits and try to keep the servers running even if the playernumbers are in the thousands on the same server. The lag will even-out the amount of players in a system and put the forces in a ratio matching each sides amount of players available. This should be a far better solution to virtual limits until CCP has come up with better coding practises, hardware or another solution. Virtual caps is bad m'kay?
16 Comments
Wow. Well thought out. Lot's of important information here. This really is becomming a problem that CCP needs to find a cure for otherwise, like you said, there's going to be lame war tactics, and I'm sure, a lot more player looking for a POS refund because the other side was using an exploit.
Thanks for the great article!
Strings
Move the servers to Linux, while your at it port the EVE client to Linux too. :-)
CCP already stated they're coming up with Linux and Mac clients by the end of summer.
I doubt moving the servers to Linux would speed anything up.
yeah, moving to linux won't help with CCP's bloated shitware.
I'd rather anything then the eventual separation of the one main server to many different ones.
Ander, while I agree that the cap isn't ideal, you can't simply say "they should have let whoever wanted to jump in do so". It WOULD have crashed the node. You say they should try to keep it working? What can they try? They could go down the data centre and waft the node with paper fans maybe, I guess they could try praying to the relevant Icelandic deities, or they could talk nicely to it and ask it not to fall over, but I'm sure you realise deep down (however much it hurts from your point of view) that the node WOULD have gone down if left un-capped. Again, and again, and again, and again.
Say again, then, how is that better than having a system cap so that people can actually play the game? Oh, wait, the attackers just keep jumping people in even though the node keeps crashing, like the coalition did against LV and eventually take down the yard with minimal losses because all the defenders in system are stuck in jump ad login queues, right? Because to say that the lag has the intelligence to even out the players on each side to approximate the forces on each side - I call BS I'm afraid. Lag is indiscriminate as you well know.
I totally accept that the system cap has just meant that the next grand strategy will be to get people in that system at a pos eve in they are afk so that they can blob the system before the other side. I realise that. It makes it very important to keep all your moons occupied with friendly POS, and it makes it very important to have covops in system probing out and popping safespotted individuals to get your side more options to jump in.
But you have to accept that there is not going to be a quick fix. We all want one but we are not going to get one. You can't rip out all the infrastructure and replace it with more powerful kit overnight even if the budget is there, and you can't just "rewrite the code" at the drop of a hat. And yet given that, everyone calls for CCP to do something. This move isn't great, but it's about the best they can do so that people can play the game, until such time as more drastic measures CAN be introduced.
Was there a battle? Yes. People shot things. The POS with perhaps a titan in production was destroyed, and the coalition leadership said they were prepared to do that at any cost. Dreads were destroyed. Pew pew was witnessed.
Where CCP dropped the ball really was in the surrounding systems from the POS because that was simply unplayable. But really, the cap is a realistic reaction to the mass warfare that we are seeing, which has clearly caught CCP on the hop, and you more than anyone, with an infrastructure of your own to support, should understand the size of the problem that they face and the fact that it's a major project to make the game handle the massive ramp-up in the scale of alliance warfare over the last 3 months.
As someone has said before, the fact we can now do 15 vs 15, 50 vs 50, 100 vs 100 in acceptable conditions is never mentioned anymore, and in a years time I suspect we will be able to play even 500 vs 500 in acceptable conditions. Whether we will ever have the opportunity to see 1500 vs 2500 at anything near playable conditions without replacing the entire cluster with a massive supercomputer, is anyone's guess.
hi bob.
Your attempt to manipulate the meta-aspects of the game in your favor, yet again (with plenty of help from ccp), will fail. System capping favors whoever has the most ships to blob up with, and the goons will win that.
Good points and I agree that crashing is not ideal either. It's hard to reinforce the servers but lets investigate this battle.
It clearly favoured the defender in either way.
1. Had there been no cap and the node had crashed the POS would've had time to recharge. This was nothing the coalition wanted.
2. Coalition had a standing order not to jump everything they had into the system. The point was to take down the POS and then gradually increase the amount of players to defend in the aftermath.
3. BoB had already got their players in. Lets says 500 , so only 200 coalitionfriendly could ever get in. While players from both sides dropped it'd have been pretty random on who could login so eventually the numbers stayed the same when BoB had capped system.
The cap is not a realistic way to solve the problem and is worse than node crashing in many cases.
Considering that node crashes will affect the defending and offending force in equal numbers that means numbers will be proportionate when node finaly comes up.
It's all about statistics. While the system cap is more about who gets in first and can keep themselves online and AFK.
In a system where there is close to the cap limit, AFK players should be dropped if they dont actively do something in a few minutes. Should help with that problem.
"Considering that node crashes will affect the defending and offending force in equal numbers that means numbers will be proportionate when node finaly comes up."
I disagree. I think JV1V proved that node crashes favour jump-ins. Let's not forget that everyone was saying "Node crashes favour jump ins and that's why bob always jumps in".
And, once all those people have jumped in, then the whole reason for the node crash in the first place (too many people) doesn't go away. There's still too many people, and it's going to continue to crash time and again when those too many people try to get back on each time the node has restarted, until people get bored and go to work/sleep/school.
I don't see *that* as a satisfactory way to determine a huge fleet battle either :)
I think what we have here are two not very good temporary solutions to a problem that can't be solved properly for some time, and we're just going to have to make the most of it by the looks of things :|
Oooh, I can just *feel* the hate flowing from you out of my screen. Would it help if I say I'm not Bob? I guess not, you won't believe me.
It is inconcievable, isn't it, that anyone other than Bob could put forward a point of view that doesn't match yours?
Seriously dude, since you know you're fighting all the GM's and all the Devs you might as well go play PotBS, so can I have your stuff?
(if I really /must/ explain, I'm joking, OK?)
Ander,
I like where you cover the facts on why this is happening. I got annoyed with the complaining about things being wrong. Do you have a solution to offer or are you just going to whine? I guess the editorializing about something that must have CCP devs in daily meetings just got to me. When you said "this is the worst tactic" I expected to see you offer a better suggestion. I was let down when you didn't.
Just to comment on that, you are correct with capping system limits, however CCP's servers cannot cope with the load of 1500 players on the same node, so the node will crash. Now if CCP choose to take the cap limit off you will find a similar tactic could be employed. If your defending a pos at the time all you have to do is get as many players online as possible so that the nodes crashes and the attackers cant do jack shit. At least with the cap limit the node wont crash and the attacking or defending players will be give a chance, however slight at accomplishing there goal. At the moment CCP are stuck between a rock and a hard place with no way out. They cant solve this problem overnight so expect a lot of capped systems and expect a lot of whining from anyone who is on the losing side to capped severs. In the end some pos's will be destroyed some wont but mostly i guess it will be a stalemate with most systems being capped with defending players rather than attackers. However Eve players will find tactics around capped servers if there simply locking the server out to attackers or planning attacks several weeks ahead and getting players to log out in systems etc etc. There is one thing for sure, someone WILL still get ganked along the way and that makes me happy :)
Seke.
Assuming that the reason they haven't done anything for the last X minutes is not just lag, then this gets the barvo seal of approval as a Good Idea (tm) ;)
if have an idea, mabey its a crappy one, and mabey some one already mentioned it above, but what if you put an alliance cap?
if an allicane cant grow bigger than 750 or so, then you cant have a massive fleet battle with 1500vs2500, becasue alliances wont be bigger than 750! altho theres probably alot of problems associated with what i said, its an idea, mabey a start
Negative.
The coalition wasn't a single alliance, it was around 10 different alliances IIRC all with the same intent to take down the titan. So that's up to 7500 on one side right there.
As for BoB - you'd just get fragmentation. An alliance that currently has 2000 members would just split up into 3 smaller alliances that shared resources and fought together.
All you could *actually* do like this is increase the annoying administration of co-ordinating the separate sub-alliances (in terms of trading resources etc).
At the end of the day if you have 3000 people in regular contact who are prepared to join together to fight for/against a cause, then you can't stop them from doing that.