Please recognise that this constantly repeated reply has been disproven. John Alexander at Aussie Broadband explained why they couldn’t fix the latency already:
And no, he isn’t just a guy at a help-desk, he’s pretty much one of their head honchos when it comes to managing their network.
@68128048 John Alexander writes…
[“Google are in Sydney and advertise it out of there. I’ve adjusted it slightly so it now goes direct to Singapore, but it looks like the return path is being preferred by Google to come back in via their PNI with us in Sydney. It’s taken 50ms off the times though.”]
You can check the looking glasses of ALL of the below, and they are all broken
Aussie Broadband
Superloop
iinet
TPG
Telstra
and every single vpn node publically available in Perth:
Perth Binarylane: Broken, 105 ping rather than 50-60
Perth RansomIT: Broken, 150 ping rather than 50-60
Perth Wombat: Broken 100 ping rather than 50-60
Perth iinet public node: Broken, 170 ping rather than 50-60.
So when you say that, you have to realise, there is no isp that isn’t broken for this issue, and those that have tried to fix it have all failed. Aussie Broadband have failed, despite it being elevated as far as it could be. Mate have failed as in the start of this thread.
There is no one who can fix it on even the isps end. It is google who has to fix how they handle their reverse routes.
The variance in latency mentioned is because players in Perth get 60-80 ms to Syd2 as well; which is entirely irrelevant to the topic of the thread, but I can see how it could confuse someone who is not as familiar with the Perth gaming landscape. Keep in mind that due to three hour time difference, and Syd2 having a small peak, playing on Syd2 is not exactly a solution.
I am focusing on Perth because that’s what I have had the most information on, but google’s terrible handling of their reverse routes also seems to be affecting India, Sri-Lanka and Mauritius.
And when you talk about Peering …
Aussie Broadband on this broken route are peering directly with google as far as I am aware.
It goes straight from Aussie Broadband to google back to Aussie Broadband?
So if peering with google causes issues with routing to google servers, doesn’t that also indicate that there’s a bit of an issue?
So when Aussie Broadband is peering with google, they hand it off to google in Sydney by default purely because even access to the Singapore server is being advertised out of there. The route stays within Google’s Network to go from Sydney to Singapore. Then back to the node in Sydney, hands back off to Aussie Broadband in Sydney, then goes back to Perth.
Even if Aussie Broadband ignores how it’s advertised and forces the forward path, Aussie Broadband hands off to google at Singapore, google in Singapore then to get to Perth decides to route via Google in Sydney, hands off to Aussie Broadband in Sydney then goes back to Perth.
This behaviour is just continually repeated. To any Australian destination, google in Singapore routes to google in Sydney, then decides how to go the rest of the way from there, regardless of isp, that is the crux of the problem. Because this behaviour has been present in literally everything I have been able to check, it’s definitely a fault on google’s end.
If you want to see why forcing all Australian routes to go via Sydney is extremely bad, then just look at a map. The distance from Perth to Sydney is only marginally less than the distance from Perth to Singapore, meaning that this broken route of forcing Google in Sydney as a checkpoint both ways causes triple the latency.
It would be akin to forcing all routes to NA west to go back and forth via New York when you live in Texas; it just doesn’t make sense when you look at a map.
And no, it doesn’t make sense to close a thread just because the problem hasn’t been fixed for 5 months.