Why I believe it is SE, and not the ISPs.Follow

I am a network engineer by profession, and have several certifications, etc. I feel this gives me a rather reasonable amount of knowledge on this subject - So no, I'm not just sitting here going "durr, it must be SE!!11"

To start, let's use Occam's Razor at it's best: Square-Enix is undergoing network congestion issues, and shortly after a large portion of the population has connection issues. It is quite logical to assume that the two are related.

It is not logical to assume that such a small update (sub100mb!) caused this issue, when such large updates as XP SP1 and SP2 were pushed out to a far greater amount of people without causing issues on the ISP end.

Now, for the more technical side:
Many people have had issues fixed by changing their IP address. Now, ISPs work by placing their customers behind a NAT (Network Adress Translation). There aren't enough IPv4 IPs out there for every single person to have their own, so you get assigned a private address, and at the ISP's border, they translate it into a public one.

ISPs take your MAC address, and use this to assign the private IP you use while still inside their network. Changing this can get you a different private IP, and thus put you in a different NAT pool.

This results in you getting a new public IP.

All of this takes place internally. You're still going to be going through the same routers to get to the internet.


But let's look at what goes on at SE's end: They have different server groups behind different routers. This can be verified by netstat and other networking tools.

They also just pushed out an update that caused them massive congestion issues.

Now, to make sure traffic flows as smoothly as possible, a thing called QoS, or Quality of Service, is used. It can allocate more bandwidth to certain protocols, or outright block traffic, and more. For those wanting to get the highest level of Cisco certification, QoS is a subject they have to learn well, and those in the industry deal with it quite often, especially in high traffic environments.

QoS is also something that can, when misconfigured, cause a networking nightmare of issues.

People getting the R0 error can still /tell other players, and people see them moving and talking, and so on. However, they are not getting NPC and other info from the servers. People are still able to send packets, which is why they can still zone (This is also reason that teleport and speed hacks work, because you report your location to the server).

Now, if QoS was adjusted to alleviate or prevent future congestion issues, then it could very well be being improperly applied to traffic. Changing your IP can get your out of a set of QoS rulings, which is why this solves the issue for quite a few people.

Now, for those that it doesn't, it is still quite simply explained: You are being placed under the same QoS rulings on this new IP address, or have been assigned another IP address under this rulings (More unlikely, due to the fact they do not appear to be handing out the QoS rules over IP ranges, as a new IP on the same class C range as my old IP worked fine for me, and the chance of being assigned the same IP as someone else under these rulings is rather slim).

As different zones are on different servers behind different routers, this explains why communication and certain still zones still work fine.



I could be wrong, as I'm certainly not positive this is the reason, and cannot be without access to SE's network, but from my experience and knowledge on the subject, this is what I believe to be the most reasonable explanation.

Edited, Mar 13th 2007 10:46pm by Exodus

"durr, it must be SE!!11"

?

what makes you think no-one else has any knowledge on the subject of networking?

my sisters boyfriend whos been here all day works in networking and he says it on SEs end, people are putting in short "it must be SE" purely because; its something that SE need to fix on their end.. not ours.

please dont be so naive. alot of people on here also know what there talking about.

Instead of giving a sarcastic retort you could just not bump it.

I find it interesting since I haven't heard this info before.


Edited, Mar 10th 2007 6:11pm by sweetumssama

It's certainly not as useful as your thread (And don't take my weighing in on several things in it as any real criticism for you - you've done a wonderful job in it), but I did feel the need to shed what light I could on the issue for those who are curious.

Edit: As for being an expert... Working on it, but I don't feel right in accepting that praise just yet ;)

Edited, Mar 10th 2007 6:13pm by Chtulhu

Oh, I didn't take it as criticism at all ^.^

When people said it was IP related, I am comfortable enough in networking to show how to obtain that and clear those settings.

The actual reason this isn't working I've been theorizing with my limited networking experience. Posts like this help to clarify the issue that is the genesis of these problems ^.^

This post wins!

you my friend, are going in my sig.

well, according to the official website, there will be a maintenance on March 12.

Probably 11 our time.... http://www.playonline.com/ff11/polnews/news9935.shtml

I called the US GM twice today and they were not helpful... gonna try the JP ones.

I was actually thinking that it may have been
a QoS issue as well. But I wasn't about to post
in those horrors of threads.

A nice, logical post.


'tis a beacon of hope amongst
the bubbling sea
of complete retardation
and logical fallacy.

This is why I believe it is QoS related. There are ways to limit and shape traffic that are triggered, and not universal. If it is improperly configured, people could be setting it off and causing this issue.

I could be totally off, but the network congestion issue is one that would likely be tackled with QoS, and a mistake in the attempt to fix that issue for the future could cause problems like people are having.

Edit: Missing a t.

Edited, Mar 10th 2007 7:29pm by Chtulhu

Brilliant hypothesis! I completely overlooked QoS as the two-edged sword. That would explain why problem IPs are different for each zone and how traffic is missing in one direction.

Not to mention, not all characters on the exact same account and server being affected.

The Google translator is quite inaccuarte, lol. Just a heads up.