Hopefully this doesn’t violate any rules or cause much friction. And TL;DR: It’s each of our ISPs, not NCSoft/Level 3.
I opened a support case for WvW Skill Lag. We all see it, at some point, often times when fighting in SMC blob vs blob. It’s something people who have been playing WvW for a lot longer than I have complained about with no resolution. I decided to open a support case on the issue. As many of you are aware, NCSoft is the hoster for GW2 and their servers are located in Texas (Dallas, if I recall correctly). This in and of itself isn’t the issue.
Level 3 is NCSoft’s ISP. This appears to be the core of the problem. My ISP is Comcast (Comcast Business class). That said, my guildies are using a variety of ISPs from all over the United States, Canada, and the world. We all experience skill lag (or outright lag) in a variety of ways. One of my guildies from overseas even only experiences lag in EBG, the map as a whole, under the right conditions. Borderlands are often lag free for this person.
But that said, ANet Support is of course our gateway to any issues regarding GW2. And this post isn’t a reflection on ANet Support’s response. They’re the voice of those who can provide a resolution when we’re discussing network or underlying service issues. ANet Support has been extremely supportive and patient looking into this and raising the problem to the right individuals within ANet/NCSoft. I do want to give them props for taking this particular ticket as far as they did; they did everything in their power to examine the issue with the data I was able to provide. Primarily, this consisted of providing rough timeframes for skill lag as well as traceroutes using their tooling of choice.
This ticket was open for almost 1.5 months and has come to a ‘conclusion’. ANet Support, per NCSoft engineers (service and/or network, I don’t personally know), have laid the blame at our ISPs, in as much as the email was almost templated as the first paragraph blamed Comcast, and subsequently went onto say “For the majority of you […] packet loss spikes are occurring at the point when Comcast routes your data onto our network, not after”. It went on to say “If you are not on Comcast, you are likely still being affected by similar problems on the part of our ISP, either right at the point when the data is being handed off from your ISP to ours, or even before then”. Again, this feels a bit templated, but I can understand why as NCSoft cannot blame one ISP as this occurs on ISPs across North America (and beyond).
Of course, the commonality between all of these users across all of these ISPs is the Level 3 hand off and of course NCSoft network/services. In fact, my WinMTR traces showed the packet loss occurring within Level 3’s network. Examples below, and they’ll be formatted poorly, so apologies in advance.
Columns are:
Resolved Address/IP | % | Send | Recv | Best | Avrg | Wrst | Last |
The percent is for packet loss percentage, while Best through Last are the round trip response time (RTT) in milliseconds.
Example #1:
| <my rough location>.comcast.net - 2 | 263 | 260 | 0 | 20 | 169 | 0 |
| 4.68.71.73 - 91 | 263 | 26 | 0 | 18 | 38 | 32 |
| 4.69.210.141 - 97 | 263 | 10 | 24192 | 24870 | 25728 | 25278 |
| 4.59.197.34 - 2 | 263 | 259 | 37 | 60 | 208 | 45 |
| 64.25.32-9.ncsoft.com - 3 | 263 | 257 | 40 | 60 | 200 | 47 |
Example #2 (the ‘no response from host’ is the Level 3 handoff):
| <my rough location>.comcast.net - 0 | 58 | 58 | 11 | 14 | 40 | 13 |
| No response from host - 100 | 58 | 0 | 0 | 0 | 0 | 0 |
| 4.69.210.141 - 99 | 58 | 1 | 24633 | 24633 | 24633 | 24633 |
| 4.59.197.34 - 2 | 58 | 57 | 65 | 74 | 164 | 67 |
| 64.25.32-9.ncsoft.com - 0 | 57 | 57 | 65 | 74 | 172 | 75 |
Both of these examples are within the Level 3 network. Not on the handoff from my ISP, not on the handoff to NCSoft. Right in the middle.
Server: Crystal Desert (so toxic!) | “Make CD DVD Again”
Guilds: [VII] – They let me claim stuff
(edited by Ni In.6578)