Your ISP has Latency Issues, and you burn all your bridges to Contact Them
When my mom and I bought a home in Seattle, I signed up with CenturyLink’s fiber-to-the-home service. To my surprise, it had latency issues initially if I run a Tor relay, even if I cap my bandwidth, and later on (meaning by now) the latency issues happen even on an idle connection, but back to the story.
I tried to contact CenturyLink through every medium imaginable: customer service, CEO emails, FCC complaints, and even legal threats via arbitration service FairShake, nothing worked.
CenturyLink either told me “our network is fine”, or suggested “solutions” like setting Quality of Service settings on my router or other solutions even when presented with PingPlotter graphs showing the issue.
Other solutions like a “second” GPON line don’t eliminate the latency spikes as I tried that and one connection’s spikes affect everyone. One tech even told me about my neighborhood having “capacity issues” which disproves CenturyLink’s “official” response.
I also got FCC complaint responses like this, and now newer FCC complaints aren’t even getting forwarded. In fact, the only way forward at the FCC now is a “formal” complaint which means hiring an attorney, while CenturyLink “stands by” running a Gigabit connection so poorly that makes AT&T’s 802.1X or Frontier’s mishaps the lesser of the two evils in comparison.
Now, the latency issues happen 24/7 so QoS won’t work if even an idle connection shows issues.
Other ISPs who also give symmetrical Gigabit service like Google Fiber’s Webpass at our old place had no issues whatsoever even if I was to max out my connection (which I never do, but still).
I did later contact someone at customer service who seemed to understand my issue and said he’ll forward it, but two weeks later the issues still persist.
I don’t know if they ignored me, or it’s the supply chain issues that are making it harder to order replacement equipment on CenturyLink’s network. But it certainly isn’t right, and I burnt all my bridges to contact CenturyLink.
When the latency spikes are extreme, it looks something like this:
On an average day now, the pings are like this:
neel@fatbox:~/misc/www/www.neelc.org % ping 63.231.10.68
PING 63.231.10.68 (63.231.10.68): 56 data bytes
64 bytes from 63.231.10.68: icmp_seq=0 ttl=254 time=4.004 ms
64 bytes from 63.231.10.68: icmp_seq=1 ttl=254 time=2.929 ms
64 bytes from 63.231.10.68: icmp_seq=2 ttl=254 time=784.931 ms
64 bytes from 63.231.10.68: icmp_seq=3 ttl=254 time=1.718 ms
64 bytes from 63.231.10.68: icmp_seq=4 ttl=254 time=2.192 ms
64 bytes from 63.231.10.68: icmp_seq=5 ttl=254 time=2.373 ms
64 bytes from 63.231.10.68: icmp_seq=6 ttl=254 time=2.226 ms
64 bytes from 63.231.10.68: icmp_seq=7 ttl=254 time=201.693 ms
64 bytes from 63.231.10.68: icmp_seq=8 ttl=254 time=2.381 ms
64 bytes from 63.231.10.68: icmp_seq=9 ttl=254 time=2.740 ms
64 bytes from 63.231.10.68: icmp_seq=10 ttl=254 time=2.020 ms
64 bytes from 63.231.10.68: icmp_seq=11 ttl=254 time=2.053 ms
64 bytes from 63.231.10.68: icmp_seq=12 ttl=254 time=1.901 ms
64 bytes from 63.231.10.68: icmp_seq=13 ttl=254 time=136.416 ms
64 bytes from 63.231.10.68: icmp_seq=14 ttl=254 time=3.065 ms
64 bytes from 63.231.10.68: icmp_seq=15 ttl=254 time=1.896 ms
64 bytes from 63.231.10.68: icmp_seq=16 ttl=254 time=3.270 ms
64 bytes from 63.231.10.68: icmp_seq=17 ttl=254 time=2.673 ms
64 bytes from 63.231.10.68: icmp_seq=18 ttl=254 time=173.981 ms
64 bytes from 63.231.10.68: icmp_seq=19 ttl=254 time=628.960 ms
64 bytes from 63.231.10.68: icmp_seq=20 ttl=254 time=2.260 ms
64 bytes from 63.231.10.68: icmp_seq=21 ttl=254 time=1.442 ms
64 bytes from 63.231.10.68: icmp_seq=22 ttl=254 time=2.679 ms
64 bytes from 63.231.10.68: icmp_seq=23 ttl=254 time=2.321 ms
64 bytes from 63.231.10.68: icmp_seq=24 ttl=254 time=1.545 ms
64 bytes from 63.231.10.68: icmp_seq=25 ttl=254 time=316.176 ms
64 bytes from 63.231.10.68: icmp_seq=26 ttl=254 time=2.610 ms
64 bytes from 63.231.10.68: icmp_seq=27 ttl=254 time=1.788 ms
64 bytes from 63.231.10.68: icmp_seq=28 ttl=254 time=10.047 ms
64 bytes from 63.231.10.68: icmp_seq=29 ttl=254 time=9.523 ms
64 bytes from 63.231.10.68: icmp_seq=30 ttl=254 time=8.507 ms
64 bytes from 63.231.10.68: icmp_seq=31 ttl=254 time=3.120 ms
64 bytes from 63.231.10.68: icmp_seq=32 ttl=254 time=20.583 ms
64 bytes from 63.231.10.68: icmp_seq=33 ttl=254 time=3.059 ms
64 bytes from 63.231.10.68: icmp_seq=34 ttl=254 time=1.848 ms
64 bytes from 63.231.10.68: icmp_seq=35 ttl=254 time=2.245 ms
64 bytes from 63.231.10.68: icmp_seq=36 ttl=254 time=11.270 ms
64 bytes from 63.231.10.68: icmp_seq=37 ttl=254 time=2.765 ms
64 bytes from 63.231.10.68: icmp_seq=38 ttl=254 time=2.093 ms
64 bytes from 63.231.10.68: icmp_seq=39 ttl=254 time=3.651 ms
^C
--- 63.231.10.68 ping statistics ---
40 packets transmitted, 40 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 1.442/59.324/784.931/162.900 ms
neel@fatbox:~/misc/www/www.neelc.org %
Why not Comcast?
Comcast’s upload speeds are very small when compared to their downloads, unless we can spend $329/mo instead of $65 we are paying now, along with a steep install fee, so it’s a no-go unless we have to.
I also don’t want Comcast’s small upload speeds, and even if Comcast did “RFoG” or “EPON” they don’t give symmetrical uploads.
Why not WaveG, Webpass, and Atlas Networks?
We are in a single-family home that these ISPs won’t service.