Skip to content

Web Scraping Proxies: Choose by Site, IP Type and Billing

Four criteria, a five-step test on your own target, and what each IP type is for.

The four things that decide which proxy a scraper needs

IP type: a datacenter IP sits in a hosting network, a residential or static ISP IP belongs to an ordinary internet provider. Some sites return the same page to both, others treat hosting networks differently, and you only find out by testing your own target.

How the IP changes: a rotating proxy gives a new IP for each new connection, while a connection you keep open (keep-alive) keeps its IP. A session or a static IP keeps one address. Independent page fetches suit rotation; anything that depends on the previous request, such as a login or a cart, needs one address.

Location: price, stock, search results and delivery options can change by country or city. If you need to see the page as a visitor from a given place sees it, pick the country (and city, where the line supports it) before you pick the price.

Billing: traffic-billed products charge per GB, so heavy pages cost more per request; per-IP products charge per IP per day, so cost depends on how many addresses you hold, not on how many pages you fetch.

What each IP type is for, and how Proxy302 bills it

What each IP type is for, and how Proxy302 bills it
ItemGood forHow the IP changesPrice (Proxy302)
Rotating datacenterOpen public pages, high volume, US exits onlyNew IP for each new connection; or one IP for about 10 minutes in a sessionUS$0.40 per GB, no prepaid pack
Rotating residential, Budget linePages that depend on the visitor's network or country; country-level targetingNew IP for each new connection; session of 1 to 120 minutesUS$1.50 per GB; 1.124 per GB in a 500 GB package
Rotating residential, Premium lineSame as above, plus state and city targeting when you generate the connectionUsually the same IP while you keep using it; it can change after it sits idleUS$5 per GB; 4.25 per GB in a 300 GB package
Static ISP, per IPA login or dashboard that must stay on one address for days or weeksDoes not rotate during the lease; 1 to 90 daysUS$0.50 for 1 day, US$7.99 for 30 days, same in every country

Prices and terms from the Proxy302 product facts, checked 14 November 2026. Per-GB lines are bought by traffic; per-IP lines are bought for 1, 3, 7, 30, 60 or 90 days. Static ISP stock differs by country and is shown on the order page.

Scenario by scenario: which one to try first

Public listings, news, catalogue pages that look the same from any network: try rotating datacenter first, because it is the lowest price per GB. If the pages come back different from what a browser shows, move to residential.

Prices, availability or search results by country: rotating residential with the country set in the connection, and a sticky session if one visit takes several requests. For city-level questions use the Premium line.

A logged-in account you read every day, or a dashboard you poll for weeks: one static ISP IP per account. Do not rotate under a login that you want to keep.

Step 1 to 5: test on your own target before you buy volume

  1. Step 1: fetch 50 pages directly

    Use your normal network and record the status codes and whether the fields you need are in the response. This is your baseline.

  2. Step 2: repeat through the cheapest type

    Rotating datacenter for pages that do not depend on location. Compare with the baseline: same status codes, same fields.

  3. Step 3: move up only if the result differs

    If pages differ from what a browser shows or the country is wrong, run the same 50 pages through rotating residential with the country set, then compare again.

  4. Step 4: confirm the exit before scaling

    Send one request through the proxy to a page that prints the exit IP, and check the country and network. Set a timeout on every request, because Requests waits indefinitely unless you set one.

    Before continuing Keep to the target site's published limits. robots.txt states a site's rules for crawlers, but per RFC 9309 it is not access authorization, so read the site's terms as well.

  5. Step 5: compare the cost per valid result

    Divide the price per GB by the number of pages that came back correct per GB. A cheaper type that returns fewer correct pages can cost more per useful page. For per-IP products, divide the daily price by the pages you fetch in a day.

Python: send one request through the proxy with a timeout

import requests

proxy = "http://LOGIN:PASSWORD@HOST:PORT"
proxies = {"http": proxy, "https": proxy}

r = requests.get(
    "https://ipinfo.io/json",
    proxies=proxies,
    timeout=(5, 30),  # connect, read
)
print(r.status_code, r.json().get("country"), r.json().get("org"))

# SOCKS5 needs: pip install "requests[socks]"
# use socks5h:// so the proxy server resolves DNS

Example only: replace LOGIN, PASSWORD, HOST and PORT with the values shown in My assets. The proxies argument and the socks5h scheme follow the Requests documentation.

Doing this on Proxy302

Rotating residential and rotating datacenter use the gateway proxy.proxy302.com: port 2222 for HTTP and 3333 for SOCKS5, with the username and password from My assets. The protocol is chosen when you generate the proxy. Traffic is bought in GB, with no free trial, so the smallest test is 1 GB of residential traffic or one static IP for one day.

Datacenter exits are US only. Residential covers 195+ countries.

Web scraping proxy questions

Do I need a residential proxy to scrape?

No. If a small test shows the same pages and fields through a datacenter IP as in a browser, a datacenter IP is enough and costs less per GB. Residential is for pages that change with the visitor's network or country.

Rotating or sticky session for scraping?

Use rotation when each request stands alone, such as a list of product URLs. Use a session when several requests belong to one visit, such as a login followed by pages. On the Budget line a session lasts 1 to 120 minutes.

Does a proxy make scraping allowed?

No. A proxy changes the network address the site sees; it does not change the site's terms or rules. RFC 9309 says robots.txt is not access authorization, so check the terms and keep your request rate reasonable.

How do I slow a Scrapy spider down?

Set DOWNLOAD_DELAY, or turn on AutoThrottle with AUTOTHROTTLE_ENABLED. AutoThrottle adjusts the delay from server latency and never goes below DOWNLOAD_DELAY.

Is there a free trial to test a scraper?

No. The smallest purchase is 1 GB of rotating residential traffic or one static ISP IP for one day (US$0.50).

Reference links

4
  1. Proxy302 product facts and price list

    Product observation · Confirmed on:

    Product facts file of the site, checked 14 November 2026: products, prices in US$ (1 PTC = US$1), lease terms, protocols and ports, quantity limits, no free trial.

  2. Requests documentation: Advanced usage, Proxies and Timeouts

    Official documentation · Confirmed on:

    Read on 14 November 2026: proxies are set with the proxies argument; SOCKS needs pip install requests[socks]; the scheme socks5 resolves DNS on the client, socks5h on the proxy server; requests do not time out unless a timeout is set.

  3. RFC 9309: Robots Exclusion Protocol

    Official documentation · Confirmed on:

    Read on 14 November 2026: robots.txt rules are crawler preferences and "not a form of access authorization"; a 4xx response for robots.txt lets a crawler access anything, a 5xx or unreachable file means assume everything is disallowed.

  4. Scrapy documentation: AutoThrottle extension

    Official documentation · Confirmed on:

    Read on 14 November 2026: AutoThrottle adjusts the delay per site from server latency, is off by default (AUTOTHROTTLE_ENABLED), aims at an average of 1.0 parallel requests per site by default (AUTOTHROTTLE_TARGET_CONCURRENCY), and never goes below DOWNLOAD_DELAY.

Ready to try it?

Create an account, top up once, then pick rotating residential, static ISP or datacenter IPs and have one running in your browser within minutes.