Planned Server Maintenance on March 30 : 08.00-12:00 GMT
zomg i carnt play game 4 four hours what are they thinkink anet gimme back my monay!!!
Heh. Just kidding. Wrote that so the idiots would not have to but I know they will anyway.
So people, I'm posting this as a notification to those who missed the network update news. Do not plan an important voyage/mission for the appointed time.
And thanks to ArenaNet for all the effort they are putting into the game. I hope they won't have a hard time with this maintenance.
Guild Wars will be unavailable during scheduled maintenance on March 30th from 00:00 PST (08:00 GMT) and 04:00 PST (12:00 GMT).
Unfortunately the Guild Wars database servers experienced a hardware failure that prevents them from operating properly, and in order to restore functionality we need to replace several components. We apologize for any inconvenience this maintenance may cause, and hope to have Guild Wars running again in short order.
For those folks with a technical bent, and for those who'd just like to know more about why we're taking such a long outage after almost a year with no scheduled maintenance, read on:
Because Guild Wars is a global game, where players from anywhere in the world can play against other gamers from any other location on the globe, there's really no good time to schedule maintenance because there isn't an idle period; there are always players from some part of the world enjoying the game. We've therefore strived to minimize the number of scheduled maintenance periods, and we've been largely successful.
While we have had several brief, unplanned outages in the nearly one year since the launch of Guild Wars, by and large we've been able to keep the game running continuously. We've had approximately twelve hours of downtime this year, ten hours of which was due to the unfortunate (but necessary) database rollback. Though the database rollback took quite a while to complete, the rollback only removed one hour of progress from character advancement, though we do realize that players lost their time, and for that we wholeheartedly apologize.
This outage is intended to allow us to perform some critical maintenance on our database infrastructure. In order to ensure that Guild Wars is always available, we purchased quite a lot of redundant server hardware (also known as clusters). The clusters are designed so that, if one part fails, another part is capable of taking over the load. Instead of one computer for a given task, there are two computers, each with multiple network cards, power-supplies, drive interface controllers, disk arrays, and all sorts of other components. While this is expensive, it's really the only way to ensure that the network continues to operate 24 hours a day, 7 days a week.
A simple analogy is that it's like having two lightbulbs (with two power supplies and two switches) to keep a room lit. If one bulb burns out, there's still time to change it before the room goes dark, unless the other bulb burns out at the same time, of course...
In the past we've had several successful cluster failovers, but unfortunately this time not everything worked as desired. Basically, one disk controller went bad, and even after replacing the controller we can't get the silly thing to come back again. To ensure that we don't have a catastrophic failure, we're taking down everything on Thursday morning (Pacific Time) so that we can replace the broken components and bring everything back to fully functioning service.
We've alloted four hours for the maintenance, and we certainly hope that it will take less time. We know that a lot of folks won't be able to play Guild Wars during that time period, but we hope that everyone will understand the necessity. We'll look forward to seeing you back in the game a little later in the day.
This part caught my attention:
Quote:
each with multiple network cards, power-supplies, drive interface controllers, disk arrays
Properly done RAID 5 ftw? D;
Last edited by koneko; Mar 29, 2006 at 12:58 PM // 12:58..
The earlier thread was deleted because there was so much misinformation being thrown around and the OP was rather vague. Claims that it'd be down the whole day, claims that it was just the support site... too many bad posts, not enough real information.
well, that's unsetling. I've worked in an environment where scheduled maintanence periods were a thing of legend, and I still remember how badly our technicians were over-worked. Um, I don't mind a little downtime from time-to-time, if it means higher play quality overall. I'll have to see exactly how improved the game is after this 4 hour period, but I'm rather unnerved about this lack of maintainence periods. =\
well, that's unsetling. I've worked in an environment where scheduled maintanence periods were a thing of legend, and I still remember how badly our technicians were over-worked. Um, I don't mind a little downtime from time-to-time, if it means higher play quality overall. I'll have to see exactly how improved the game is after this 4 hour period, but I'm rather unnerved about this lack of maintainence periods. =\
It's that whole nightmare all over again.
Dont understand what you mean there? Do you want regular scheduled maintenance instead of the 12 hours that were unscheduled? I know what I want... the 12 hours that are unscheduled. Whats half a day!!!
I'll have to see exactly how improved the game is after this 4 hour period, but I'm rather unnerved about this lack of maintainence periods. =\
It's that whole nightmare all over again.
Game won't be improved _that much_. You won't probably see or feel anything different after the maintenance. They'll just be replacing some hardware, which failed, and backupsystems didn't start properly. Current systems might be suffering from higher-than-usual-load, but they are working. Once the maintenance is complete, loads go back down to normal level on all the servers they have.
You dont need that many maintenance periods for the whole network, if you build it to be fault-tolerant, as A.net has done. One part fails, other, similar parts pick up the slack.
Redundancy ftw!
Quote:
Originally Posted by Huntmaster
WoW has what, 12 hours of downtime a day?
Totally different server-architecture. GW is built around instances, WoW is build around huge persistent world, which requires (or so I've been told) reboots, to clear all the little memoryleaks, open databases and so on.. Then again, it might just be crappy programming.
Closest thing to GW-system is DDO, which is built around non-combat areas and heavily instanced, same as GW. They have weekly maintenance, though
Last edited by Ogg; Mar 29, 2006 at 11:07 AM // 11:07..
You wouldn't want that for a database. In that case performance is everything. So RAID 1 FTW (for the redundancy in one box and to survive single disk failures) and distribute it among as many channels and disk arrays as you can...
You wouldn't want that for a database. In that case performance is everything. So RAID 1 FTW (for the redundancy in one box and to survive single disk failures) and distribute it among as many channels and disk arrays as you can...
*Buzz!!* Guess again. RAID 0+1. Performance + redundancy. I know a little something about clusters having built and maintained a couple hundred of them.
*Buzz!!* Guess again. RAID 0+1. Performance + redundancy. I know a little something about clusters having built and maintained a couple hundred of them.
This is getting a bit off-topic, but do you really mean what you say?
RAID 0 doesn't give you any performance boost, as the number of disk writes/reads on a given number of disks and a given Bus System (be it SCSI, FC or whatever) is limited. And when running high volume databases (as the Guild Wars DB probably is) you'll want to get as many individual disks on as many different channels as possible.
RAID 0 does nothing to the number of read/writes your disks can serve, it just gives you a different view of the disks used.
Let's do some maths. Assuming one disk can handle 1000 reads / writes per second (just to have a number for simplicity). Putting a number of n of these in a RAID 0 would give you n*1000 reads / writes per second. Using n individual disks gives you the exact same n*1000 reads / writes per second.
If this were "I need to write a constant data-stream with high speed to some disks", RAID 0 would be a nice choice, but I've never seen an online database create that kind of load on a disk
Yeah Havelock, we don't want to get off topic, but RAID 0+1 gives you added performance with redundancy. Since this is their DB and it's constantly being tapped, they'd need high performance in addition to redundancy. It's been pretty much the standard for DB clusters for a few years now. If I were to guess, I'd bet they're running an Oracle DB on Sun or HP high end servers connected to an EMC. But that's pure speculation on my part.
Yeah Havelock, we don't want to get off topic, but RAID 0+1 gives you added performance with redundancy.
It seems you really believe that. So be it. But as I stated above: to double the performance (even with RAID 0) you have to double the number of disks.
I'll admit: with 4 disks forming a RAID 0+1 you get more performance (and more capacity) than with 2 disks forming a RAID 1 but what else could you expect?
Thank god someone finally linked the actual post about this. Now I can see why they are taking it offline instead of using their fail overs to run parallel units while they work on one. Oh well looks like I gotta get a life for four hours tomorrow. that kinda sucks. LoL