Summary of an article published on The Web Scraping Club on 2026-04-26 (free to read). Read the full article
Past a few thousand requests a day, the bottleneck is egress and IP reputation. WAFs rate-limit and fingerprint, so a single IP is dead in minutes. Datacenter proxies get flagged by ASN and CIDR block, and residential proxies are billed by bandwidth, which gets expensive at scale. The question we start from is whether you can scrape at scale without burning budget on proxies.
We look at NyxProxy, a self-hosted HTTP/SOCKS5 proxy server that uses the IPv6 /64 subnet many cloud providers hand out with a VPS. We explain how the address space works, why ndppd and NDP proxying are needed for responses to come back, and how the tool keeps a pool of addresses and rotates them. Then we set it up on a VPS and check it against an IP echo API from Python with Requests. We also say plainly why this is not a silver bullet and list the limits that keep commercial proxies in business.
What’s inside
- How a /64 splits into network prefix and host part, and why that gives you 2^64 usable addresses
- What happens behind a residential proxy gateway, and why buying proxies means delegating your whole egress infrastructure
- The three rotation parameters (pool size, max usage per IP, max age) and what each one does against WAF tracking
- Four limits: IPv4-only targets, bans on the whole /64, datacenter ASN reputation, and nf_conntrack exhaustion on a small VPS
In the full article: The quick-setup command, the config.yaml rotation settings, the expected daemon output, the Python test script with its sample results and the server requirements checklist.