Hello,
Just today, I've noted very severe packet loss issues in my conncetion, resulting in unacceptable (> 1000ms) latency in any sort of online gaming. This issue is not limited to me or my own connection ; all gamers that I know connecting to the same server who live in the Valley area, and those who live in the Valley only, are experiencing the same night-and-day decline in performance.
I believe I have identified the source of the sudden problems, but a bit of background is needed. About a month ago, persistent issues arose from Cox's use of Cognet-hosted servers in order to route my signal to, amongst other places, the World of Warcraft server to which I connect in Chicago. Long story short, those servers, for whatever reason, are associated with poor performance and a lot of latency. After multiple calls, Cox rerouted the path to this server through hops that did not appear to be Cognet-hosted. This change immediately cleared up the persistent latency issues for not only myself but also for several other people living in the Valley area who had been reporting similar issues.
Just today (9 January), my tracerts revealed that Cox has swapped back to Cognet servers. The extreme latency and packet loss have immediately returned with this change, for both myself and other gamers that I have spoken to who live in the Valley. Only the inital and terminal hops on the Cognet network are the same IP's as they were a month ago. Unfortunately, the first hop, 154.54.13.129, happens to be the server that was known to be causing issues the last time around.
I know that there are Cox technicians who browse these boards, and I'd like to know if possible why Cox swapped back to the terrible Cognet servers in the first place. Additionally, I'd like this change to be reverted, because at present, the poor connection precludes any sort of actual gaming. There are no suitable ISP alternatives in my area.
The following image displays the acceptable trace route at top, taken from a PingPlotter sample recorded yesterday evening, and the Cognet trace route at bottom from tonight. I have chosen trace routes to my Chicago-based World of Warcraft server because I know my signal is routed through the Cognet network to get there. I am experiencing no issues with connecting to servers that do not entail routing my signal through said network.
[att=1]
If a comparison would be needed, here is an old traceroute from 4 December when similar lag issues were occuring prior to my swap off the Cognet servers. Cognet hops that I have used previously are bolded.
1 NELTHARION [192.168.1.1]
2 10.100.64.1
3 70.169.77.212
4 chndcorc01-te-0-4-0-2.ph.ph.cox.net [70.169.72.52]
5 chnddsrj01-ae2.0.rd.ph.cox.net [70.169.76.229]
6 langbprj02-ae2.rd.la.cox.net [68.1.1.19]
7 te7-5.ccr02.lax05.atlas.cogentco.com [154.54.13.129]
8 te0-1-0-1.ccr22.lax01.atlas.cogentco.com [130.117.50.89]
9 te0-3-0-6.ccr22.iah01.atlas.cogentco.com [154.54.3.185]
10 te0-4-0-6.ccr22.dfw01.atlas.cogentco.com [154.54.0.206]
11 te0-2-0-6.ccr22.mci01.atlas.cogentco.com [154.54.2.230]
12 te0-5-0-4.ccr22.ord01.atlas.cogentco.com [154.54.45.150]
13 te0-5-0-3.ccr22.ord03.atlas.cogentco.com [154.54.43.234]
14 att.ord03.atlas.cogentco.com [154.54.12.86]
15 cr1.cgcil.ip.att.net [12.122.84.54]
16 gar2.clboh.ip.att.net [12.122.133.197]
17 12.122.251.22
18 63.240.130.210
Note that even in the best circumstances, the Cognet route would have an extra hop, resulting in unneccessary latency. The top image corresponds with a ping of about 77ms to the server, whereas the bottom one results in a ping of 900-1200ms. The apparent packet loss after 63.240.130.210 is the result of pings past that hop being blocked and is not indicative of a problem.
I also find it interesting that the Cox server 68.1.1.19 shows consistent 40-50 % packet loss with PingPlotter, regardless of my actual connection performance. I have never gotten less than 40 % packet loss at this address at any time of day or on any connection. This particular hop does not appear to affect my connection quality at all, but as I have never recorded any instance where there was not packet loss at this address, I have little basis for comparison.
For reference for any Cox tech that may come accross this thread, I am presently on the Ultimate package. I can provide further details if it will help bring about a resolution of this problem.
↧