I'm wondering if anyone can explain this phenomena I'm experiencing.
About once every week or two, for about 2-3 hours (usually during the morning), my RTTs drop by about 25ms. You can see this here via my attached smokeping (between the hours of 10AM and 12:15PM). After the time elapses, my RTTs go back up.
Here is a traceroute ran during the lower latency:
Tracing route to google.com [74.125.228.226]over a maximum of 30 hops: 1 1 ms 1 ms 1 ms RT-00259C5E6014 [192.168.1.1] 2 15 ms 15 ms 15 ms 162.104.240.1 3 15 ms 14 ms 15 ms 71-32-28-17.rcmt.qwest.net [71.32.28.17] 4 22 ms 21 ms 21 ms dcx2-edge-02.inet.qwest.net [67.14.28.138] 5 22 ms 22 ms 22 ms 65.122.30.210 6 22 ms 22 ms 34 ms 209.85.252.80 7 22 ms 22 ms 22 ms 72.14.233.95 8 22 ms 22 ms 22 ms iad23s24-in-f2.1e100.net [74.125.228.226] Trace complete.
Here is the same traceroute ran right after the latency increased:
Tracing route to google.com [74.125.228.99]over a maximum of 30 hops: 1 1 ms 1 ms 1 ms RT-00259C5E6014 [192.168.1.1] 2 34 ms 35 ms 35 ms 162.104.240.1 3 34 ms 34 ms 34 ms 71-32-28-17.rcmt.qwest.net [71.32.28.17] 4 41 ms 42 ms 41 ms dcx2-edge-02.inet.qwest.net [67.14.28.138] 5 41 ms 42 ms 41 ms 65.122.30.210 6 42 ms 42 ms 42 ms 209.85.252.80 7 42 ms 42 ms 42 ms 72.14.238.253 8 42 ms 41 ms 41 ms iad23s08-in-f3.1e100.net [74.125.228.99] Trace complete.
My packets are taking an identical path, but what's with the discrepancy? Obviously, I'd prefer the lower latency, but I have no idea what's the cause of the fluctuations or if I'm powerless to do anything about it. The fact that it usually occurs in the morning wouldn't lead me to believe it's a load issue. I remoted into a family member's PC who is in another county about 25 miles away, and verified the same results (their latency dropped the same time mine did, and rose the same time mine did), so it wouldn't be a local DSLAM issue. In fact, here are their traceroutes:
During latency drop:
Tracing route to google.com [74.125.228.227]over a maximum of 30 hops: 1 <1 ms <1 ms <1 ms DD-WRT [192.168.1.1] 2 <1 ms <1 ms <1 ms PK5001Z.PK5001Z [192.168.0.1] 3 24 ms 27 ms 24 ms nc-71-48-208-1.dhcp.embarqhsd.net [71.48.208.1] 4 24 ms 40 ms 23 ms 71-32-28-113.rcmt.qwest.net [71.32.28.113] 5 31 ms 31 ms 31 ms dcx2-edge-02.inet.qwest.net [67.14.28.138] 6 31 ms 31 ms 31 ms 65.122.30.210 7 31 ms 31 ms 31 ms 209.85.252.80 8 31 ms 31 ms 31 ms 72.14.233.95 9 31 ms 31 ms 31 ms iad23s24-in-f3.1e100.net [74.125.228.227] Trace complete.
After latency increase:
Tracing route to google.com [74.125.228.233]over a maximum of 30 hops: 1 <1 ms <1 ms <1 ms DD-WRT [192.168.1.1] 2 <1 ms <1 ms <1 ms PK5001Z.PK5001Z [192.168.0.1] 3 44 ms 43 ms 43 ms nc-71-48-208-1.dhcp.embarqhsd.net [71.48.208.1] 4 64 ms 43 ms 43 ms 71-32-28-113.rcmt.qwest.net [71.32.28.113] 5 50 ms 61 ms 51 ms dcx2-edge-02.inet.qwest.net [67.14.28.138] 6 50 ms 52 ms 51 ms 65.122.30.210 7 51 ms 54 ms 51 ms 209.85.252.80 8 51 ms 51 ms 50 ms 72.14.233.95 9 51 ms 51 ms 50 ms iad23s24-in-f9.1e100.net [74.125.228.233] Trace complete.
Sorry for the long post - wanted to make sure I covered all of the bases. From what I can tell, the latency has to be added by a layer 2 device in between our premise equipment and the first hop since it's not appearing on the traceroute, but that's purely speculation. Any ideas?
↧