I spoke with IT specialist from us i give him every info what i have.
his expression
Anyway, this looks at the AT&T issue in HKIX, because the latency has grown most of the way through given hop, and the subsequent hopes have been with increased latency.
Sometimes it is obvious from a given hope that high latency is the setting of the given router who is doing peering (it is only a matter of ICMP being served by the main CPU and not ASICs), but on the next hop is the latency OK.
Imho has nothing to do with VPS.
The problem might have been either the last equinex hope in the HKIX or ATT hope line in HKIX, and it is also dependent on the router being physically in HKIX and peer-to-peer.
Imho pure network problem, because when you look at the latency of 3 last hopes so constantly 34 / 35ms
If there was a problem on the IB side, it would seem that the increase in latency was right there ... and now, with 4ms latency.
Otherwise, it does not make sense if traffic goes to HKIX.
Because every DC wants to keep DC traffic in DC ... do not let him out to keep the capacity for the data going out of DC or DC.
And so I hope that the problem wil not appear ... oftentimes someone makes a mistake but does not want to admit and that's how they got lost until it is solved so miraculously.
