Santiago VPS: my month-long experiment

I needed a South American presence for a monitoring project and leased a 2GB HostHatch instance in SCL for 30 days. Here are my measurements, since the community debates this endlessly.

Routing to GRU (São Paulo): consistently poor. 180-220ms via Miami, occasionally via LAX. No direct peering observed; HostHatch's upstream in SCL appears to lack local Brazilian transit. BGP communities not honored on prepend requests—I set 3x prepend to force Cogent, no effect. RPKI validation shows their IRR entries stale, last updated March 2025.

Routing to MIA: 110-130ms, acceptable. MIA-NAP latency stable, no packet loss in 86400 sample.

The premium over Miami: 40% for identical specs. Unless you need the IP reputation in LACNIC space or have specific compliance requirements, the edge location is not justified by performance. The routing table does not care about your feelings.

One observation: my Chilean bank flagged my US cards less from this IP. YMMV on whether that's worth the surcharge.

GreenCloudVPS LAX: $7/GB, routes to GRU via... madrid?? 340ms
HostHatch LAX: $9/GB, your LA path
RackNerd LAX: $5/GB, actually decent to GRU at 45ms but RAM is oversold to hell

• Banking access: worth $4/month premium?
• Latency to Brazil: only RackNerd delivers
• My spreadsheet says: LA + VPN for banks, save the diff

Actually, the BGP community issue (which (if you check HostHatch's looking glass — https://bgp.he.net)) is that they filter standard communities on ingress. You need (extended communities (specifically (target:64512:100))) to shift traffic. Most providers (actually (all the ones I've tested in SCL)) do this (for (some (reason))).

I had a Santiago box with RackNerd for eighteen months and it was fine until it wasn't and the day it died I was eating avocado toast which I know is cliche but it was good avocado and my cat jumped on the keyboard and I thought she broke something but actually the router in their SCL facility had a power supply failure and I only found out because I was checking my monitoring dashboard which I check every hour even when I sleep and the weather that day was rainy which is rare for Santiago in February and I kept refreshing and refreshing and their status page said "degraded performance" for six hours before they admitted outage and when I finally got ticket response they said "we are aware" and I said I was aware too and they didn't think that was funny and my cat didn't care and I moved to Miami after that and never looked back and people who say edge locations are worth it haven't lived through the SCL dust storm of 2024 which killed three datacenters and also my neighbor's garden

To be honest, I used OVHcloud's SCL pop for two years. As said, they shut it down in January 2026.

• Power: 2N UPS, but grid instability 47 days/year
• Transit: $12/Mbps to Brazil vs $2/Mbps to Miami
• Customers: 340 active, 12 needed GRU routing

The economics don't work. They sold the IP space to HostHatch. If you need Chilean banking access, rent a residential proxy instead. Cheaper, better compliance, less headache.