Montreal hamsters are tired, SE. Fix or breed new ones, please.

[Lemuel81]

snip

sadly enough she is right. A GPN is needed for a lot of people. The problem lies in our ISP not so much with SE. Pay for a GPN is cheap or youll have to find another game to play
When I moved to another ISP, during raid times I have to use a GPN at times.

[Malzian]

Okay, so let's talk about the Internet, routing and ISPs for a moment (before anyone asks, I have a degree in Networking and Network security, so I assure you I am qualified to speak on this topic).

First to dispel a major misconception (for those of you knowledgeable what the Internet actually is, you can skip this section) - the Internet isn't just one solid amorphous blob out there where information just sits waiting for you to go and get it, it's a series of computer networks that are interconnected together that have to take your request and sort through to find out on exactly which system in which network the information you are looking for is. This, in itself, can take a little bit to actually work out though mostly it's set up so this process takes a millisecond or so. Once your request has been successfully parsed out and your destination has been identified then your request is transferred from your PC to the system that is holding the information which will then work to fulfill your request be it retrieving the file to load a website, perform some calculations or other operations for you and then send the information back to you along a similar (note, not always the same) route in which it took to get to that system. This exchange can be a short path, or it can be an unnecessarily long path, and the longer the path the longer it takes for the information to move along it.

Onto ISPs and backbone carriers. Most people only ever think of their internet as being the service provider you connect to. This can be as varied as ComCast, Time Warner, Verizon, Optimum or Bob's Discount Internetz (don't know if that actually exists, but you get the idea). However, these ISPs are actually not the main connections of the Internet and only serve as the end-point linkup for consumers. The major ISPs/Telecoms rely on companies and infrastructure systems known as Backbones in order to actually send information around the world. You can think of this literally like your own nervous system where the ISPs and your connections are the nerves that connect your spine to your muscles and the spinal cord is the transfer system that does most of the heavy lifting, hence the term 'Backbone'. Level3 is one of these main internet Backbones and serves a lot of the ISPs in the Eastern/Central timezone.

Now neither these Backbone companies nor the ISPs are non-profit organizations and they each require a fee in order to transmit data across their systems. For the user-end ISPs this comes in the form of your bill where you pay money each month to access their service. However, the way Backbones and ISPs conduct business is a little different. Whenever an ISP wants to link in to a Backbone system the two companies will enter into a working contract, usually this involves terms of equity in data flow where the levels of information transmitted from ISP to Backbone is even with the amount of information received by the ISP from the Backbone along with some basic rate fees for the connection itself. This connection point and agreement is where things can get a little bit hairy since typically these through-puts have a limited amount of data they can handle before requests are made to wait which is where you end up getting your increased ping and lag from.

Now, for anyone who doesn't know what the term 'Net Neutrality' actually means in terms of this transfer, basically the concept is that if I have two pieces of information flowing through my system... say one that's a Google search and one that's a Netflix stream... I can't actively discriminate between the two items and make the Netflix data go through my network slower than the Google search data. It also means that I can't stop one particular type of data entering my network as opposed to another, nor can I charge fees for one service's data to move through my system any faster than any other. However, I am allowed to look at the data that comes to my system to determine whether or not that data is part of a legitimate exchange or if it's an invalid request such as in the type used in a DDoS attack. The consequence of this is that the larger the amount of data I need to look at the slower the process becomes (think of Airport security, when the line is long it seems to take each person much longer to get through than if the line is short).

All in all, Level3 has shown itself to be relatively paranoid when it comes to the data entering into their systems and it seems to be a major bottleneck for most services that need to run along it. Sometimes, for some reason, it seems as if the system will just arbitrarily start rejecting data of a specific type and I'm guessing it's probably something to do with their DDoS detection being set a liiiittle too sensitive. Of course, it's really just theory and conjecture since it's not something you can really prove outright, but it has all the hallmarks of it. Sadly there's little the consumer can do aside from contacting their ISP with the Traceroute information at hand showing that it's the Level3 juncture that's causing the problem, and even then it's not something your ISP can just flip a switch and fix. Really the best you can do is use something like BattlePing or WTFast to get around the suspect network leg, but even that has its limitations and drawbacks as well.

Also, this is true as well:

The problem lies with Level3, which routes between your ISP and the ISP that SE uses. It has to do with the hardware - Level3 really needs to upgrade the data center that handles packet switching between the US and Montreal, but they're hoping that the US cable providers will give them money to do so. The switches can't handle all the packets they're bombarded with, and buckle under the stress. This results in packet loss, which means some of the data between you and SE gets dropped, and you get lag in exchange because your client has to second it twice for the data to even go through - if it goes through at all. ("Rubber banding" is what happens when the data is lost to the ether permanently.)

Basically, it's a giant game of chicken, because Level3 oversold their data capacity to the US ISPs, and now they don't want to pay for the upgrades needed to meet their obligations.

[Twilite]

I think comcast has been throttling it lately.

If so isn't that against the law now?

I've read about that, though I can't say for sure it's a thing now, but I hope that's the case.

[Catwho]

Alright. I'm willing to be convinced. But -- tell me why the issues do not occur on the Level3 hops in Montreal, but only at SE's end. (For what it's worth, probably not much, my ISP claims that problem is clearly at the source, and not with any hops in between... so no help there.)

The problem lies with Level3, which routes between your ISP and the ISP that SE uses. It has to do with the hardware - Level3 really needs to upgrade the data center that handles packet switching between the US and Montreal, but they're hoping that the US cable providers will give them money to do so. The switches can't handle all the packets they're bombarded with, and buckle under the stress. This results in packet loss, which means some of the data between you and SE gets dropped, and you get lag in exchange because your client has to send it twice for the data to even go through - if it goes through at all. ("Rubber banding" is what happens when the data is lost to the ether permanently.)

Basically, it's a giant game of chicken, because Level3 oversold their data capacity to the US ISPs, and now they don't want to pay for the upgrades needed to meet their obligations.

Edit: Malzian's post about their DDoS detection being set too sensitively is actually quite plausible as well, and makes sense. Thanks for the insight!

[Malzian]

Edit: Malzian's post about their DDoS detection being set too sensitively is actually quite plausible as well, and makes sense. Thanks for the insight!

Happy to help. -=^.^=-

[Crissy]

Level3 really needs to upgrade the data center.... The switches can't handle all the packets they're bombarded with, and buckle under the stress. This results in packet loss, which means some of the data between you and SE gets dropped, and you get lag in exchange because your client has to send it twice for the data to even go through - if it goes through at all.

Edit: Malzian's post about their DDoS detection being set too sensitively is actually quite plausible as well, and makes sense.

I appreciate the response -- this explains the remarkable consistency of the problem; last night, for example, it was like a switch: all loss and errors stopped at precisely 10:54 pm EST. I'm assuming the reason why packet loss and errors shows up on SE servers and not on the next series of hops with Level3 is because loss/errors recorded by tracerts are reflected one or two hops from their point of origin.

I'll investigate a GPN, again.

[fortunavex]

"no lag issues in Canada" I beg to differ. If anything, plagued with 90,001 and 90,002 errors. My husband plays on PS4 and I play on PC and we're constantly dc'ing at different times. He dc's more than I do on PS4 which is really weird. Why 2 people on the same internet, in the same house, never dc at the same is a funny and hilarious problem in the middle of savage! It's the absolutely worst on Sundays around 9:15 pm ATL time. Go Go Rogers.

I'd ask who your ISP is, but sadly I only know what Roger's Atlantic was like (and 2 years ago at that). I'm on Cogeco in Ontario and never had a problem aside from their shitty hardware giving bad wifi signals. Keep in mind though that DSL is affected by distance from the closest hub, and that cable/broadband is affected both by that and how many people are using the network locally to you too. If you're in one of the big cities out East, might be the reason. I'd talk to your ISP if it affects more than just FFXIV.

[Nihility]

for me the horrible lag starts about the same time every day but it's definitely not comcast
MTR diagnostic shows the massive packet loss once it jumps to level3, sometimes it's even 80%+

It sucks to have to pay for a virtual private network but I don't think there is any way to manually force your connection to avoid level3 and it fixes the problem completely

You can try using a different DNS server like googles but there is no guarantee it will change your connection route

te-11-0-2.edge4.Frankfurt1.level3.net (4.68.127.105) is the particular jump where it goes all to hell for me

[Malzian]

for me the horrible lag starts about the same time every day but it's definitely not comcast
MTR diagnostic shows the massive packet loss once it jumps to level3, sometimes it's even 80%+

It sucks to have to pay for a virtual private network but I don't think there is any way to manually force your connection to avoid level3 and it fixes the problem completely

You can try using a different DNS server like googles but there is no guarantee it will change your connection route

te-11-0-2.edge4.Frankfurt1.level3.net (4.68.127.105) is the particular jump where it goes all to hell for me

Just want to highlight this since people might not notice it in there... te-11-0-2.edge4.Frankfurt1.level3.net (4.68.127.105)

[Catwho]

Looky what I found.

customercare@level3.com.

Email your tracerts there, guys. If enough of us spam their CS address, maybe they'll actually pay a little attention.