Skip to content

Bandwidth accounting includes incoming

VPS Hosting by bellaauc 24 replies 3.6K views
#11
Mahdi861 said:
You cut off there mate. Throttled or charged?

Charged. Hetzner bills EUR 1.19/TB overage, both directions. I hit 847GB over my 20TB plan, so only 0.8TB, but at 3AM when the bill arrived it looked like a lot more. My fault for the truncated post—forum timed out.

Updated checklist:
- Check if your provider counts both directions
- Check if "unmetered" has a hidden cap
- Check if overage is throttled or billed
- Check if overage billing is pre-pay or post-pay (Hetzner is post, surprise on invoice — docs: https://docs.hetzner.com)

3-2-1 or you're already dead
#12
bellaauc said:
Hetzner bills EUR 1.19/TB overage

Same in Berlin. Their Cloud Console shows a "bandwidth" graph that looks like total traffic. You have to click through to the project settings to see egress vs ingress broken out. Dark pattern? Maybe. German efficiency? Also maybe.

#13

I moved my backups to rsync.net. ZFS snapshots, no incoming metering because they price by allocated storage not transfer. More expensive per GB but predictable.

Still keep a Hetzner box for compute. Just don't back up to it anymore.

my cloud. my rules. my 3AM alerts.
#14
SamAlvi said:
rsync.net

We evaluated them at work. Good for targeted backup, bad for "I need a full VPS to run tests on my restores." Bella's use case is valid. We ended up with two providers: Hetzner for compute, Wasabi for storage. Wasabi has egress fees though, different trap.

Four person platform team, we have spreadsheets for this now.

#15
nina64 said:
Wasabi has egress fees

Wasabi also has minimum retention. 90 days. Delete early, pay anyway. Another TOS trap.

I self-custody on a Mini PC in Medellín now. 200Mbps symmetric, no accounting. Backup to friends in Bogotá and Cali over WireGuard — https://www.wireguard.com. Not 3-2-1 strictly but I control the keys.

not your keys, not your coins
#16

Hetzner's Helsinki datacenter gave me worse incoming ratios than Falkenstein. Same plan. No idea why, maybe routing asymmetry. I moved the box to Nuremberg, problem halved. Not solved, halved.

#17

This thread saved me. Was about to push 2TB initial seed to a new Hetzner box. Doing it to RackNerd LA instead.

#18
astro18 said:
Hetzner's Helsinki datacenter gave me worse incoming ratios than Falkenstein.

Routing asymmetry would affect path, not ratio. More likely your source peers differently with Hetzner's upstream in Finland vs Germany. Check your traceroutes.

Or Hetzner allocates different upstream capacity per location and shapes differently. No way to verify without internal data.

-t

worst bandwidth, best stories
#19

From Osaka: Hetzner Singapore incoming is brutal. 300ms to my home ISP, backups crawl, incoming piles up. OVHcloud Sydney was worse, their dashboard counters lag by 48 hours. You find out Monday what you did Friday.

3 #20
carlos2 said:
Mini PC in Medellín

Same energy. J'ai un vieux Dell Optiplex à Lyon, 500/500 fibre Orange, pas de limite. Backups vers chez mon frère à Grenoble. On est deux nœuds, pas un cloud.

Mais pour le public, j'ai encore besoin d'un VPS. OVHcloud Gravelines, je savais pas qu'ils comptaient l'entrant. Facture de janvier, surprise.

Post a reply

You need an account to reply. Log in or register to join the conversation.

Post reply Preview Save draft