Web scraping hosting
Answer
Scraping infrastructure needs addresses that are not already blocked and capacity to run headless browsers. Incognito VPS allocates from clean IPv4 pools checked against 60+ blocklists, sells additional addresses at $2.50 per month from separate /24s, and does not require identity verification.
| Scenario | Minimum |
|---|---|
| HTTP scraping | 2 vCPU, 4 GB RAM |
| Headless browsers (5–10) | 6 vCPU, 16 GB RAM |
| Large crawl + parsing | 16 vCPU, 64 GB RAM |
| Address diversity | $2.50/mo per extra IPv4 |
Address reputation decides whether this works
Most anti-bot systems check the address before they check anything else. Ranges belonging to well-known clouds are pre-classified as datacentre traffic and treated with suspicion by default; ranges with a history of abuse are simply blocked.
Our IPv4 pool is checked against sixty-plus blocklists at allocation and every six hours after, and the acceptable-use policy that removes abusive customers is what keeps it that way. Additional addresses come from separate /24s so a single block does not take out your whole pool.
Sizing for headless browsers
Plain HTTP scraping is cheap — a small VPS will saturate its network long before its CPU. Headless Chrome is the opposite: each instance wants roughly 300 to 500 MB of RAM and a meaningful share of a core, so concurrency is bounded by memory.
Ten concurrent Chrome instances is comfortable on 16 GB. If you are running Playwright or Puppeteer at scale, budget memory first and cores second.
Do it properly
Respect robots.txt, rate-limit yourself, identify your crawler in the user agent with a contact address, and cache aggressively so you are not re-fetching what you already have. This is not moralising — it is how you avoid being blocked, and the operators who get blocked are the ones who hammer.
Our acceptable-use policy prohibits credential stuffing, scraping behind authentication you are not authorised to hold, and collecting personal data in breach of applicable law. Ordinary public-web data collection is fine and always has been.
Frequently asked questions
01 Are your IPs blocked by anti-bot services?
Our ranges are not on abuse blocklists, which is the main thing you control. Any datacentre range can still be classified as datacentre traffic by a sophisticated anti-bot system — that is inherent to hosting, not specific to us.
02 Can I get multiple IP addresses?
Yes, $2.50/month each, up to sixteen per server, allocated from separate /24s so one block does not take out your whole pool. No justification form.
03 Which plan for headless browsers?
BASTION (6 vCPU, 16 GB) runs about ten concurrent Chrome instances. Each instance wants 300–500 MB, so memory is the binding constraint, not cores.
04 Is scraping allowed under your AUP?
Public-web data collection is fine. Credential stuffing, accessing data behind authentication you are not authorised for, and collecting personal data in breach of applicable law are not.
Related