While I cannot confirm officially, there is a lot of rumors, that several larger UK ISP's are throttling traffic at that time period. I am not sure who to contact, but the individual ISP's to solve this, from your point, maybe another NANOG'er knows.
Hi Lasse, Thanks for the reply. We wrote an app to reveal troubles. Just to satisfy any curiosity and get some facts out, I will provide a real world example (1 of many) from a direct test of one of our BT sourced customers (this is from a 08-29 test at ~22:04 hours GMT): Date IP RTT Port ActualRecv NicSent NicRecv 090829 22:04:24 86.128.0.0/10 103.5 80 199 13 214 090829 22:04:24 86.128.0.0/10 103.5 80 199 13 214 090829 22:04:24 86.128.0.0/10 103.5 443 200 12 215 090829 22:04:24 86.128.0.0/10 103.5 443 199 12 214 090829 22:04:24 86.128.0.0/10 103.5 7255 2 2 5 090829 22:04:24 86.128.0.0/10 103.5 7255 3 1 4 090829 22:04:24 86.128.0.0/10 103.5 7003 3 2 5 090829 22:04:24 86.128.0.0/10 103.5 7003 4 1 5 090829 22:04:24 86.128.0.0/10 103.5 7202 27 3 32 090829 22:04:24 86.128.0.0/10 103.5 7202 24 2 29 090829 22:04:24 86.128.0.0/10 103.5 7499 27 3 32 090829 22:04:24 86.128.0.0/10 103.5 7499 25 2 31 090829 22:04:24 86.128.0.0/10 103.5 80 195 13 206 Idle NIC bandwidth Send: 0 KB/sec Recv: 0 KB/sec To remove any doubt we also measured idle bandwidth utilization on the NIC when the test wasn't run to remove any other culprit such as torrent download, A/V streaming and etc in the background. In this case, 0/0 on idle use. All results are in KBytes I withheld the actual IP address of this test and replaced it from the source prefix. We have quite a few iterations of similar results from other source addresses from this prefix alone. All appear to exhibit the same issue. I've already written British Telecom and they never replied.