SamAlvi
Member
OP
Self-Host Everything
- Joined:
- Jun 2024
- Posts:
- 188
- From:
- Portland, US
Contabo wanted €84 to renew my 2GB VPS. Same specs were €48 in their BF email blast. I opened a ticket and flat-out threatened to migrate everything to OVHcloud if they wouldn't match. Took three back-and-forths but they caved. Renewal locked at $48/yr.
Why pay for managed reverse proxy when you can self-host it with docker compose and a Caddy container? That's what runs on this box. Three compose stacks, total RAM footprint under 400MB. The savings from this negotiation cover my secondary DNS at RackNerd with room left over.
Anyway, the lesson is: actually follow through on the threat. I had the OVHcloud cart ready and screenshot attached. They could smell it.
https://www.ovhcloud.com/en/vps/
my cloud. my rules. my 3AM alerts.
pam3
Member
Latency Witch
- Joined:
- Aug 2024
- Posts:
- 74
- From:
- Melbourne, Australia
Threatened to migrate everything to OVHcloud
Contabo observed 14ms additional latency to OVHcloud's Limburg presence versus their own, per observed traceroute [1]. RFC 792 suggests ICMP echo alone does not capture application-layer performance [2]. The price differential of $36/yr equates to $0.0986/day; amortized over observed 8760 hours, the break-even threshold for migration effort remains unclear.
100% packet loss at hop 3
larryjeong
Member
- Joined:
- Jun 2024
- Posts:
- 179
- From:
- Texas, US
Sam. You know OVHcloud is just. Contabo's budget subsidiary. Right. They share the same ASN upstream. Same racks in two of three locations. You threatened to leave for their own company. And they still negotiated against themselves. That is. Something else...
$3/year. 128MB RAM. Pure happiness.
SamAlvi
Member
OP
Self-Host Everything
- Joined:
- Jun 2024
- Posts:
- 188
- From:
- Portland, US
You threatened to leave for their own company
I did not know that and I do not care. The cart was real, the screenshot was real, the ASN lookup I did not do. If Contabo wants to negotiate against themselves that is their business. My business is paying $48 not $84.
The Caddy setup is three containers: caddy, crowdsec, and a custom fail2ban sidecar. 400MB includes all of it. I have seen people run heavier Electron apps.
my cloud. my rules. my 3AM alerts.
larryjeong
Member
- Joined:
- Jun 2024
- Posts:
- 179
- From:
- Texas, US
Now you do. But seriously check peeringdb. Same upstream for 40% of routes. The other 60% is Cogent and you know how that goes.
I have a Contabo box in St. Louis and an OVHcloud one in Vint Hill. Both hit 200ms to my house in Dallas because Texas fiber is a joke. The latency between them is 6ms. Same metro, same problems.
$3/year. 128MB RAM. Pure happiness.
pam3
Member
Latency Witch
- Joined:
- Aug 2024
- Posts:
- 74
- From:
- Melbourne, Australia
Your sample methodology is unspecified. Single-origin MTR or multi-path? Time of day variation? LINX congestion events correlate with UTC 14:00-18:00 [1]. Without longitudinal data the stddev is uninterpretable.
I maintain probes to both from three vantage points: Melbourne, Sydney via AARNet, and Perth via IX. OVHcloud Sydney consistently outperforms Contabo Singapore for AU eyeballs by 8-15ms. The $36/yr is noise against transit costs if your users are here.
100% packet loss at hop 3
