YuriDavid
Member
- Joined:
- May 2024
- Posts:
- 174
- From:
- Kyiv, Ukraine
---
- routing reality
- submarine cables exist
- ASC: direct
- AJC: japan loop
- marketing claims
- "optimized path"
- WARNING: may mean "we have a POP in SYD"
- NOTE: POP != cable ownership
---
- actual topology matters more than latency promises
indentation is not optional
Hana5
Member
GPU poor
- Joined:
- Jun 2024
- Posts:
- 104
- From:
- Tokyo, Japan
Please check server routing table. GreenCloudVPS say "optimized" but help route go to US west coast.
I use HostHatch instead, direct path very much.
8-char NTLM found in 4min 32sec
tomchan
Member
- Joined:
- Jul 2024
- Posts:
- 202
- From:
- Bristol, UK
Fun little project: built my own anycast with NAT VPS nodes, it actually works
# bird config snippet for local pref
protocol bgp {
local as 64512;
neighbor 10.0.0.1 as 64513;
import filter {
if (net ~ [ 43.250.0.0/16+ ]) then {
bgp_local_pref = 300;
}
};
}
Still cheaper than premium routing claims
works on my bench ¯\_(ツ)_/¯
GeorgeNmp
Member
AS64512
- Joined:
- May 2024
- Posts:
- 218
- From:
- Ashburn, US
The Asia Submarine-cable Express (ASE) and Australia-Singapore Cable (ASC) both offer direct paths. What matters is whether your provider purchases transit from someone with actual cable access versus reselling Tier 1 routes that default to Pacific routing.
BGP communities worth checking: 2914:420 (NTT domestic preference) or similar from your transit. RPKI validation status of the prefixes involved. Some providers announce more-specifics over suboptimal paths to avoid peering costs.
RackNerd specifically: their "optimized" product page references latency figures that imply direct routing, but their IRR entries show AS-SET membership with networks that lack ASC landing rights. Marketing fiction, as suspected.
iBGP, eBGP, don't care, just peer