Morden Kain.3489

Correct, and all of that is inside AWS. Everything starting with 52.93.249.25 and after is AWS. Hence why I keep saying AWS has a routing problem, as they drop a lot of packets, even for the routers that are responding to TCP, UDP, and ICMP requests.

Here are a couple more from today for a server, I was in Tangled Depths at the time:

UDP Trace Route (100 "pings" per line)...


TCP Trace Route, I actually make it to the server, at four different levels (routing, gotta love it!) (100 "pings" per line)...


And for S-n-Gs, here is the topology showing where the server was found...


As I keep saying, there is a problem with the routing within AWS...

@"Inculpatus cedo.9234" ... Why do you always say that? It is not correct, as you will not get 100% packet loss. It all depends on the type of trace route you do. Yes, ICMP is the most common way to do a trace route. There are programs that do allow you to choose TCP or UDP as well. What is your response to this:


@"Feirlista Xv.1425" this is not just a simple network problem. This is something that has been with this game for nearly 8 years. Originally, when the game was hosted by NCSoft, the ping times (or lag) was bearable. Even then, my ping times were around 80ms on Comcast. When you got into large groups, you got to a point where only the auto attack would work (the 1 key normally). Then, the servers would catch up with the network load and *boom*, you had super speed.

Since going to AWS, ANet has had nothing but congestion problems with the network traffic. So, that leads to one of two possibilities (or both). First option is that ANet is using crap network code. The second option is that AWS has problems with the amount of traffic from MMOs. Looking at the internals of AWS using TCP, UDP, and ICMP, you will find that AWS has a tendency to drop packets. You do not believe me, pick up you favorite trace route tool that allows for those three communications and do a trace route to the IP address of your WvW/PvE/PvP server and let all of us complaining about the lag of 1000-4000ms know what your results are. This is *very* much a real problem with Guild Wars 2 (or as some are now calling it Lag Wars 2).

The backbone that everyone in the Pacific NW uses is Level 3 Networks, owned by CenturyLink. Now, in my particular case, I got north to Seattle, then to CenturyLink HQ in Kansas City. Apparently, the entry point for AWS is there for the servers I have hit in the past (otherwise no clue why this is). The TCP trace route shows good connection until I hit the Edge-19 routers that CenturyLink likes to use (this adds 50ms to my pings). After that, I bounce to one more router in CenturyLinks network, then to AWS. Once inside AWS, I start seeing packet drops. See the following image from 9 Apr 2020 (and is still a problem today, as it has not changed much). The trace route was done a total of 10 times, with 3 pings to each hop.

![Visual Trace Rt 2020/04/09](https://i.imgur.com/mFm630P.png"")

For the above, in game I was experiencing 2-4k ms ping times within WvW. My normal pings are between 80-120ms no matter the zone or the number of people, so this is not my system being bogged down by graphics. Before you ask, here are my system specs:

CPU: Ryzen 7 gen 1

GPU: nVidia 2070 8GB

RAM: 16GB

HDD: NVMe 256GB

Net speed: 60Mbps down, 5Mbps up with QoS enabled at 90% of up and down speeds

EDIT: I was incorrect on the number of pings, I have corrected it to the proper number from 10. This means each hop was ping-ed a total of 30 times. My bad.

@"Ayrilana.1396", while I agree to an extent with your statement about a player's setup, my experience with this GW2 lag is not pleasant. It affects both WvW (some servers are worse than others) and PvE with high ping times and skill lags (this has gotten worse since March 2020). If you want to investigate the problem just do a trace route to the server you are experience the lag with (do a /ip within GW2 to get the IP address). In every case, when I do a trace route, the problems are within AWS. AWS is dropping packets, as some of my ICMP, TCP, and UDP traces have over 80% packet loss *within* AWS. I wish ANet would look at their Service Level Agreement with Amazon, and figure this stuff out, as AWS is not providing the service that is needed.

@"rooney.7968" , while I do agree with you when it comes to it being a ANet problem, as this is mainly within AWS, ANet does have a Service Level Agreement (SLA) with AWS. That SLA does allow ANet certain lee-ways to have AWS improve their network performance, otherwise a lawyer somewhere is not doing their job. The SLA should have stipulations on network reliability. If ANets customers are notifying them of a network problem, then ANet needs to contact AWS about the problem to have them fix the problem per the SLA.

Right now, after this last week, WvW has gotten better, as the server was changed out. I am not sure why this would be, as I have not done a trace on the new IP address. Something tells me there may still be packet loss within AWS.

@"thepenmonster.3621" I too am on CL, and doing a Visual Traceroute to the IP address listed for WvW, I get lots of packet loss within the Amazon owned AWS. In addition to this, I also have a spike of 50ms going from the West Coast to Kansas City (where CL HQ is located). CLs cer-edge-19 router adds that extra 50ms. While I normally would chalk this up as distance, this is within their own network, and going even to Level 3 Networks in Seattle (another CL owned company), I do not get these large spikes up.

When I have done the traceroutes, I normally use UDP and TCP, not ICMP (ICMP is what is normally used when using tracert). With both of those, have been doing this off and on the past few days, I normally get packet loss starting in the AWS, not within CL's network. For myself, the packet loss starts with the IP 150.222.240.61, and gets worse (up to 80%+ packet loss) at 52.93.129.130, 150.222.243.201, & 52.93.28.116. This is after 10 traces with 3 tests per hop, and a total of 32 hops using "Whatsup Visual Traceroute"with the latest version of WinPCap so I can use the TCP/UDP traceroute features. All are free (will need to sign up for crap email from ipswitch.

If you want to get results, tell CL that you are trying to telecommute using RDP, and that the lag is horrid.


Since the US went to a lot of telecommuting, there has been lot of lag. WvW has gotten intolerable. When I did a trace route using ICMP, UDP and TCP, AWS was dropping a lot of UDP & TCP packets at the following internal AWS IPs: 150.222.242.166 (33% packet loss), 52.93.129.130 (93.33% packet loss), 150.222.242.124 (80% packet loss), 52.93.28.108 (86.67% packet loss), 52.93.28.74 (40% packet loss), 52.93.28.104 (80% packet loss). The mapping was all over the place within AWS, as Amazon was trying to route the traffic around the problem.

This is most certainly a problem within AWS, as all the above IP addresses belong to Amazon. ANet, if you have your eyes on these forums, you will want to communicate with Amazon about their crap internal network.

EDIT: I was tracing to 35.168.81.98, the IP given for WvW.