You need more infrastructure, SE

[BLM]

This is a senior management issue, as was the failure of 1.0 They can get fucked. They need to let Yoshi do whatever he needs to do...

[NefarioCall]

They can a hundred servers, but the legacy world's will still be congested. Pretty sure that's the issue.

1.0 sold 600,000 copies.
That's a crap ton of potential players.

[zenmetsu]

The issue is that each world should be a cluster of servers, and the application should be executed in parallel on multiple machines, preferably VMs, with a handful running each zone. If a zone starts getting congested, they spin up additional VMs and handle the load. It isn't rocket science.

If each world is truly a single server, then epic lulz is what we have.

[Alhanelem]

Obviously, having to merge servers isn't a great thing, but it's better than having all the servers jammed and turning people off.

[Griffster]

SE is attempting to use a small funnel to control a stream coming out of a fire hose.

At least all the pissed off people come Tuesday-Saturday next week will drop the game.

Shame it will come to that. I was hoping that they would have planned their infrastructure better.

Maybe allow entire LS/FC to transfer to a JP world to alleviate the issue.

Then maybe we can log in.

[Pumra]

Them not knowing this many people were going to playing isn't a valid excuse. They gave out the # of beta keys. They gave out certain # of EA codes from preorders. They sold whatever amount of copies of the game to all the retailers. Did someone not keep track of all those numbers? Well apparently not. The # of people playing is not a an excuse for now day MMOs. This isn't their first game, it's not their first MMO, MMO is not a new genre of gaming any more. I expect to see problems...but crashing and maintenance several times to fix a problem that theres to many people playing and its still not fixed....mind blown.

[zenmetsu]

Amazing... it is surely a cold day in hell. I got a message that the JP Group list couldn't be updated, and then I got right in.

... and then the game crashed to desktop for the first time since Beta1. Awesome progress mates. Keep up the good work.

[Xeonerable]

They should have have taken the error 3102 epidemic that happened on open beta seriously and investigated if there would be any other potential server errors. They shouldn't just be now testing their servers for traffic stress, that is what open beta is for. It seems those 3 days didn't cut it.

I think I would have preferred the servers stayed in japan so the dev. team can look at it directly. I'd rather deal with a slight lag than constant errors, log-in issues, and frequent maintenance that doesn't seem to solve anything.

[Rizon]

1 obvious fact must be recognized. The concept of multiple Data Centers id is a complete and utter failure. This is especially true is some of those centers are not under SEs direct control. If SE cannot arrange to have additional servers added for each world in the next day or two then it should make arrangements to consolidate everything in 1 place, even if it is in Japan. If they do not take immediate steps to rectify this then they will not have to worry about congestion in the future as they will not have enough players to pack the worlds they now have.

[zenmetsu]

1 obvious fact must be recognized. The concept of multiple Data Centers id is a complete and utter failure. This is especially true is some of those centers are not under SEs direct control.

You really have no idea what you are talking about.

IDCs such as Equinix have staff on-hand that will perform tasks on-site. No need to have your own staff on site.

Secondly, most of the work can be done remotely if the servers are virtualized.

Thirdly, there is this thing called georedundancy. Virtualization packages like VMWare allow for transferring VMs from a failed host to a different DC if the storage (SAN, etc) is being backed up to the remote DC. If the DC gets wiped out by earthquake, nuclear blast, asteroid, etc... you can keep on playing your MMO as long as SE had the foresight to implement an effective disaster recovery plan. Then again, it's just a game.

Lastly, few corporations own DCs; they almost always lease rack space and a pipe.