Resources

A scraping VPS: bandwidth and addresses, priced honestly

Scraping is bound by two things: how many distinct addresses you can present, and whether the bandwidth line on the invoice is predictable. A metered cloud makes the second one impossible to forecast, which is why most crawlers end up on a VPS.

Most chosenIV

Work

6 vCPU · 16 GB · 200 GB NVMe

€10.40/mo

12-month term, billed once, VAT-free outside the EU

Configure

Frankfurt: 2 ms
Amsterdam: 2 ms
New York: 3 ms
Singapore: 2 ms

Why people move here

REV. D

VPS for scraping and data collection

Nine addresses on one instance

One dedicated IPv4 is included and eight more can be added at €1.20 each per month, from the region's own prefix. A routed /48 of IPv6 is €2, which is a very large number of source addresses for sites that speak v6.

Traffic with a ceiling, not a meter

Outbound is allowanced per plan — 12 TB on Work, 40 TB on Monolith — and inbound is never counted. Over the allowance the port shapes rather than billing you. No egress invoice has ever left this company.

Region diversity at one price

Geo-fenced targets need a source address in the right country. Thirty-one countries, all at the same rate, means a crawler fleet can match its targets instead of its budget.

Where the line sits

Collecting public data is fine. Credential stuffing, brute forcing, vulnerability scanning against third parties and anything that amounts to an attack is not, and it is terminated without refund. Rate-limit politely and nobody will ever contact you.

Full specification table

REV. D

Summary

Recommended plan: Work — 6 vCPU, 16 GB, 12 TB traffic
Heavy pipelines: Forge — 8 vCPU, 32 GB, 20 TB, 10 Gbit/s
Addresses: 1 + up to 8 IPv4, optional routed /48 IPv6
Egress billing: None — allowance then shaping

31 countries. 42 sites. One flat price.

Operating system

Docker Engine on Debian 13

Engine, compose plugin, and a hardened daemon.json. Nothing else.

Fleet

REV. D

Compare every plan

TERM

CORESDedicated core

Longer term, lower monthly rate

PART EVPS-BASEMost chosen

III

Base

€5.50/mo

billed €66.00 oncesave 20% vs 6-month

vCPU: 4 × EPYC
Memory: 8 GB DDR5
NVMe storage: 100 GB NVMe
Traffic: 8 TB
Port: 2.5 Gbit/s

Configure

PART EVPS-WORKREV. D

Work

€10.40/mo

billed €124.80 oncesave 19% vs 6-month

vCPU: 6 × EPYC
Memory: 16 GB DDR5
NVMe storage: 200 GB NVMe
Traffic: 12 TB
Port: 2.5 Gbit/s

Configure

PART EVPS-FORGEREV. D

Forge

€18.90/mo

billed €226.80 oncesave 21% vs 6-month

vCPU: 8 × EPYC
Memory: 32 GB DDR5
NVMe storage: 400 GB NVMe
Traffic: 20 TB
Port: 10 Gbit/s

Configure

PART EVPS-ANVILLow stock

Anvil

€35.90/mo

billed €430.80 oncesave 20% vs 6-month

vCPU: 12 × EPYC
Memory: 64 GB DDR5
NVMe storage: 800 GB NVMe
Traffic: 30 TB
Port: 10 Gbit/s

Configure

Questions

REV. D

Frequently asked questions

01How many IP addresses do I get?

One dedicated IPv4 and a routed /64 of IPv6 with every plan. Up to eight extra IPv4 addresses can be added at €1.20 each per month, assigned from the region's own prefix so geolocation stays consistent. A routed /48 of IPv6 is €2 a month if you are running your own network behind us.

02No KYC — so anything goes?

No. No identity checks means we do not care who you are; it does not mean we do not care what leaves our network. Phishing, spam, malware distribution, botnet command-and-control, credential stuffing, outbound attacks and material involving minors get the instance terminated immediately, without refund and without appeal. The acceptable use policy is two pages, written in plain language, and we apply all of it.

03What happens if someone files an abuse report about my server?

We forward it to you with the evidence and give you 24 hours to respond, except for active outbound attacks and CSAM, which are cut immediately. We do not suspend on receipt of an unverified complaint, and we do not hand over customer data without an order valid in the jurisdiction the server sits in. When we do receive such an order, we tell you unless we are legally barred from doing so.

04Which region should I pick?

The one closest to your users, not to you. Every location costs the same, so there is no reason to compromise. The locations page lists median round trip from residential probes in each metro, and there is a live latency test on every site page. If you are unsure, Frankfurt and Amsterdam reach the most people in the fewest milliseconds.