It seems the routes chosen by large Australian ISPs into the Australian regions is a little bit... off. Not sure where responsibility lies here, but something definitely needs to be done.
For the Australia SE region (i.e. Melbourne), traffic from Melbourne travels up to Sydney before entering the Azure network and coming back down to Melbourne. So rather than say 5 ms of latency, we currently sit on 28 ms (7 ms to Sydney + 7 ms to Melbourne+ 7 ms to Sydney + 7 ms to Melbourne).
Here's a traceroute from the largest ISP in Australia, Telstra, from Melbourne into an Australia SE address:
1 gigabitethernet3-3.exi2.melbourne.telstra.net (203.50.77.53) 0.332 ms 0.224 ms 0.245 ms 2 bundle-ether3-100.win-core1.melbourne.telstra.net (203.50.80.129) 1.243 ms 1.612 ms 2.120 ms 3 bundle-ether12.ken-core10.sydney.telstra.net (203.50.11.122) 14.486 ms 14.106 ms 14.487 ms 4 bundle-ether1.ken39.sydney.telstra.net (203.50.6.146) 16.736 ms 21.100 ms 23.982 ms 5 micros19.lnk.telstra.net (165.228.136.34) 14.610 ms 14.852 ms 14.883 ms 6 xe-1-2-2-0.mel01-96cbe-1b.ntwk.msn.net (191.234.81.59) 27.844 ms 25.959 ms 26.105 ms
Here's a traceroute from the second largest ISP in Australia, iiNet, from Melbourne into an Australia SE address:
1 gi4-5.dr1.mel8.on.ii.net (150.101.212.246) 80 msec 4 msec 0 msec 2 xe-1-2-1.cr1.mel8.on.ii.net (150.101.34.32) 12 msec 12 msec 12 msec 3 ae1.cr1.cbr1.on.ii.net (150.101.33.24) 12 msec 8 msec 8 msec 4 ae0.cr1.cbr2.on.ii.net (150.101.33.7) 8 msec 8 msec 8 msec 5 ae2.br1.syd4.on.ii.net (150.101.33.22) 20 msec 12 msec 12 msec 6 as12076.nsw.ix.asn.au (218.100.52.4) 12 msec 12 msec 12 msec 7 xe-1-2-2-0.mel01-96cbe-1b.ntwk.msn.net (191.234.81.59) 24 msec 24 msec 28 msec