Originally Posted by
moondog
I suggest you advise your wife to do a limited speed test (full test is and just save the top 20 results is also okay if she has time). To sort, she can export to CSV, open in excel, and the use the text to columns wizard. I usually sort by ping, but sometimes, I make a new index column that combines ping with bandwidth (not really necessary because it's easy to eyeball the results; fast ping with low bandwidth is not a winner). It's important that she has this data in hand before contacting support because those guys have no way of knowing the best performing servers are from her specific connection (they just know what works best from China as a whole).
My wife this morning (PDT) and last night (around 10-11 pm China time) was having issues with all VPNs. "A" wasn't even allowing her to login so she couldn't even get to the step of trying different servers. I figured out one of the IP addresses it uses for the login process and both of us ran a traceroute and ping to it. I was getting latencies of around 110-120ms while she was seeing about 95-99% packet loss. Packets were getting through but so many of them were being dropped it was basically an unusable connection. AC could form a connection but basically wouldn't transfer packets, etc.
Why would there be 95-99% packet loss? Bandwidth oversale? Her ISP is China Telecom, are they particularly bad about this, or can you pay more to get priority during congestion? Although if you look at the traceroute, the first node that started having major packet loss was after a CMCC node so I'm trying to figure out who is dropping the ball here.
Around 11 pm local time, packet loss dropped dramatically and she was able to login to "A" and try different servers. Japan 3 was still having issues but she was able to connect to Santa Clara...but then that connection dropped after 5 minutes
.
So I am trying to find out the answer to the question:
who is dropping the ball? China Telecom? CMCC?