Proxies for scraping become necessary the moment your request volume exceeds a site's per-IP limit: the source starts returning captchas, HTTP 429 or empty pages. Spreading requests across a pool of addresses lets you collect public data — prices, catalogs, search results, listings — in dozens of threads without downtime.
ProxyFarm offers three options for this: rotating proxies from 1 490 ₽/mo (≈ $16), IPv6 from 12 ₽ per IP (≈ $0.13) and residential proxies in 48 countries from 290 ₽/GB (≈ $3). All types work over HTTP(S) and SOCKS5, and access is delivered automatically 2 minutes after payment.
Pricing for scraping proxies
HTTPS and SOCKS5 on one port, unlimited traffic on datacenter and mobile proxies, free IP replacement within 24 hours. 15 % off for a quarter, 25 % off for a year.
How websites throttle scrapers
Most sites count requests per IP per minute or hour. Once the threshold is crossed, the address gets a captcha, a delayed response or a temporary block. Data-center subnets are tracked separately: if an IP belongs to a hosting provider, its limit is usually lower.
Hence two requirements for data-collection proxies: a pool large enough that each IP handles few requests, and the right address type for the specific source. Simple sites are fine with data-center IPv4 and IPv6; protected platforms need residential or mobile IPs.
The third requirement is speed. A 50-thread scraper with a 10-second timeout will sit idle on slow addresses. When testing, aim for a response time under 400 ms: that is normal for data-center IPv4 and IPv6 in Russian data centers.
Rotating proxies for scraping: one port, new IP per request
The easiest option to integrate is rotating proxies. The scraper connects to one address and port, and every new request leaves from a different IP in the pool. No need to maintain an address list, check availability or weed out banned IPs — we do that on our side.
Plans from 1 490 ₽/mo. Works with Scrapy, A-Parser and your own Python or Node.js scripts: a single proxy in the settings, and the code stays the same as the pool grows.
The limitation of rotation is the lack of a persistent session. If the source requires login or keeps state in cookies, a pool of dedicated IPv4 or residential proxies in sticky mode (IP held for up to 30 minutes) is the better fit.
IPv6 and residential: when to use them
IPv6 proxies from 12 ₽ per IP are the cheapest way to build a large pool: a thousand addresses costs several times less than a hundred IPv4. They suit sources that support IPv6: Yandex (Russia's leading search engine), Google, major international services, many marketplaces. Check that the target site opens over IPv6 before buying.
Residential proxies are needed when a site aggressively cuts data-center subnets: airline tickets, hotel booking, classifieds, stores with anti-bot protection. Traffic goes through ISP home addresses in 48 countries with city-level targeting, from 290 ₽/GB. You pay for volume, so it makes sense to disable image and script loading on residential.
Dedicated IPv4 from 79 ₽ per IP are used for authenticated scraping, when every account needs a permanent address.
Proxies for scraping: matching the type to the source
Pick the proxy type according to the specific site's protection, so you do not overpay for residential traffic where IPv6 is enough.
| Data source | Proxy type | Price from |
|---|---|---|
| Yandex and Google SERPs, suggestions | IPv6 or rotating proxies | 12 ₽ per IP, 1 490 ₽/mo |
| Marketplaces: prices, stock, product cards | Residential or mobile | 290 ₽/GB, 490 ₽/mo |
| Classifieds, aggregators | Rotating proxies + mobile | 1 490 ₽/mo |
| Sites with login, personal accounts | Dedicated IPv4 | 79 ₽ per IP |
| Geo-restricted international sites | Residential, city-level targeting | 290 ₽/GB |
| Unprotected sites, high volume | Shared IPv4 | 35 ₽ per IP |
Rental for 1, 7, 30 and 90 days. A daily plan suits one-off projects; for ongoing monitoring, a quarter with a 15 % discount.
Setup in A-Parser, Scrapy and Key Collector
In A-Parser, the proxy list is loaded in the Proxy Checkers section: enter addresses as host:port:login:password, and the tool checks them itself and drops dead ones. For rotating proxies, add a single line and set the thread count in the task settings.
In Scrapy, the proxy is passed via the proxy meta parameter of a request or through middleware such as scrapy-rotating-proxies. Set CONCURRENT_REQUESTS to the number of addresses and DOWNLOAD_TIMEOUT around 10 seconds so a slow response does not hold a thread.
Key Collector (a popular Russian SEO keyword tool) takes a proxy list in its network settings and spreads requests across it automatically. A sensible load for Yandex is one thread per 5–10 addresses with pauses between requests. Pull the address list through the dashboard API and verify it before launch on the proxy checker page.
Rotating IPv6 for BAS and multi-threaded scrapers
If your targets support IPv6 (Google, YouTube, Facebook, most international platforms), the cheapest option for multi-threaded scraping is a gateway to a dedicated /48 or /36 block. One proxy in the project settings: the user-rotate login gives a new address on every connection, while user-session-{{thread_id}} pins an address to a thread. Each thread works from its own /64 subnet, so blocking one address does not affect the rest.
A shared rotating IPv6 pool costs 1 490 ₽ a month; a dedicated /48 with rotation and sticky sessions is 9 000 ₽, a /36 is 20 000 ₽; traffic is unlimited. Setup guides for BAS and ZennoPoster are on the IPv6 proxies page.
FAQ
How many proxies do I need for scraping?
It depends on the source's limit and your volume. Rule of thumb: if a site allows about 60 requests an hour per IP and you need 6 000 pages an hour, you need roughly 100 addresses. With IPv6 that is cheap, and rotating proxies solve it with a single port.
What is better for data collection — rotation or a list of IPv4?
Rotation is simpler to integrate and needs no pool management: one address in the scraper settings. A list of dedicated IPv4 is needed when a persistent session matters — login, cart, cookies. Many teams combine both.
Do scraping proxies work with Yandex without captchas?
Captchas depend on the request rate per IP, not just the address type. At one thread per 5–10 addresses with pauses between requests, data-center IPv4 and IPv6 return results reliably. If captchas still appear, enlarge the pool or lower the rate.
Can I fetch the proxy list automatically?
Yes, the dashboard has an API: the scraper pulls the current address list before launch, and after an IP replacement the list updates without manual edits.
What if some addresses do not work with the target site?
Contact support within the first 24 hours and we will replace them free. Every IP is checked against spam databases before delivery, but some sites block whole subnets, so replacement for a specific platform is part of the service.