Best Rotating and Residential Proxies for Web Scraping

A rotating proxy hands your scraper a different exit IP address for each request, for a run of requests, or for a session of a set length. The site on the other end sees traffic spread across many addresses rather than a stream from one.

This page covers what rotating and residential proxies are, what separates one provider from another, and where each of the services below stands today. Connecting a proxy in code, testing it and rotating it from a script are covered in the guide to proxies for web scraping.

What are rotating and residential proxies?

A rotating proxy server switches the IP address it presents to the target. Instead of one static address used for every request, it gives a new address after each request, after a certain number of requests, or when a set amount of time has passed.

Rotating proxies come in three types. Residential proxies use IP addresses assigned by internet service providers to real households, which is why sites treat them differently from server ranges. They are slower than datacenter addresses and are usually sold by the gigabyte.

Datacenter proxies use addresses that belong to hosting providers. They are faster and cheaper, and they are also the easiest for a site to recognize, because whole ranges are registered to known hosting companies.

Mobile proxies use addresses from mobile carriers, which makes them the most expensive of the three. They are worth the price only where the data is not reachable any other way, such as content served exclusively to mobile apps.

Why use rotating proxies for web scraping?

A site that counts requests per address sees a pool as many visitors rather than one. A crawl spread across a pool therefore runs into HTTP 429 responses later than the same crawl from a single address, and the rate limit that would have stopped it applies to each exit separately.

Pools are usually spread across countries, which matters when the page itself changes by region. Prices, availability, and Google search results differ from one country to another, so a scraper that needs the German version of a page asks for a German exit.

What Separates One Rotating Proxy Provider From Another

Providers differ less on marketing claims than on what the pool is made of, how much control you get over rotation, where the addresses are, how many connections you may open at once, and what you are billed for. The rest of this section takes those in turn.

Rotation control

Two rotation models cover most of the market. Per-request rotation gives every request a new exit, which suits independent page fetches such as a list of product URLs. Sticky sessions hold one address for a few minutes, which is what a multi-step flow needs, because a login, a filter and a paginated result set fall apart when the address changes in the middle. Providers that offer both let you choose per endpoint, and the ones that only rotate per request rule out anything with a session.

Speed and reliability

Residential addresses route through consumer connections, so they answer more slowly than datacenter addresses and fail more often. A provider’s own success-rate figure is measured on its own targets, so treat it as a claim rather than a measurement and run a trial against the sites you actually scrape.

Geolocation

If the data is location-specific, such as local search results or real estate listings, the pool has to hold addresses in the places you need. Country-level targeting is standard. City-level and ISP-level targeting are not, and on several networks they sit on a higher plan.

Concurrency

Plans cap how many requests may be in flight at once, and that cap decides how long a crawl takes far more directly than the size of the pool. A hundred thousand pages at ten concurrent requests is a different project from the same hundred thousand at two hundred, so check the limit on the plan you are pricing rather than the one on the enterprise tier.

Billing model

Residential pools are usually billed by traffic, so a page that weighs two megabytes costs the same whether the scrape succeeds or the site returns a challenge page. Request-based billing shifts that risk, because a failed request is either free or cheap. Port-based plans charge for a number of simultaneous connections instead, which suits a steady crawl more than a burst. The model matters more than the headline rate, since the same job can cost twice as much on one and half as much on the other.

Security and privacy

A proxy provider sees every request you send through it. Read what the provider says about logging and about where its residential addresses come from, because a pool assembled without informed consent is a legal problem you inherit.

Customer support

When a pool starts returning challenge pages at three in the morning, the difference between a provider with a support channel that answers and one without is the difference between a delayed run and a dead weekend.

Top rotating proxy providers

We have already written about the best free proxies for web scraping. The services below are the paid ones worth knowing, with what each sells and where it fits.

HasData

HasData is a web scraping API platform rather than a proxy seller, with a dashboard and no-code scrapers next to the API. A request goes to the API, the API sends it through its own datacenter or residential exits, renders the page when you ask for rendering, and returns the HTML or parsed JSON. Rotation, retries and the browser are on its side, and billing is per successful request.

That last line is the practical difference from a proxy subscription. You do not configure a proxy in your HTTP client at all, and you cannot point your own browser at it either.

Oxylabs

Oxylabs sells residential, datacenter, mobile and static ISP proxies, with rotating and sticky sessions on the residential pool. Targeting goes down to city level, and plans are sold by traffic, with per-port options on some products.

It is the safe answer when a project needs every pool type from one account and the budget is not the deciding factor.

Proxyrack

Proxyrack offers residential, datacenter and mobile proxies with rotating endpoints. Its pricing page sells access by plan rather than by gigabyte, which makes a monthly bill easier to predict than a traffic plan on a crawl whose page weight you cannot control.

Predictable billing is the reason to look at it, and the reason to test first is that the pool behind those plans is not described in public.

Bright Data

Bright Data runs residential, datacenter, mobile and ISP networks with city-level targeting, sold pay-as-you-go or on a monthly plan. It carries every pool type in this list, which is convenient for a project that needs two of them and heavy going for one that needs a single endpoint.

Pay-as-you-go is the part that matters for a trial, because it lets you find out whether residential addresses solve your problem before you commit to a plan anywhere.

Decodo, formerly Smartproxy

Smartproxy renamed itself Decodo, and smartproxy.com now redirects to decodo.com. The product line did not change shape. Residential, datacenter, mobile and ISP pools, rotating or sticky, sold by traffic, with scraping APIs billed per request next to the raw proxies.

The rename is the thing to remember when you search for documentation, because most of what is written about this provider still says Smartproxy.

Shifter

Shifter sells residential and static ISP proxies, billed by traffic on the proxy plans and per request on the scraping APIs, with unlimited-bandwidth options on parts of the range.

Unlimited bandwidth changes the arithmetic for jobs that pull heavy pages, where a traffic plan charges for every megabyte of images you did not want.

Zyte

Zyte folded Smart Proxy Manager into Zyte API, and the pricing page now lists Zyte API alone. The proxy behaviour survived the move, which is why the old Smart Proxy Manager page still exists and points at the new product. Requests go through the API, it picks and rotates the exit, and you pay for successful responses.

If you find an older tutorial that configures Smart Proxy Manager as a proxy endpoint, expect the setup to have moved rather than the capability to have gone.

StormProxies

StormProxies sells backconnect rotating proxies, dedicated rotating IPs and residential rotating proxies with unlimited bandwidth. Its own site answers a plain HTTP request with a challenge page and renders normally in a browser, which is worth knowing before you script anything against its dashboard.

It fits a single steady crawl rather than a project that has to target a city or an ISP, and the money-back window is worth using deliberately rather than letting it lapse.

Two names that used to sit on this list are gone. The domain netnut.io no longer serves the company’s site and shows a seizure notice from the US Federal Bureau of Investigation instead, and geosurf.com does not resolve at all.

Best practices when using rotating proxies

A few habits save more time than any provider choice.

1. Test a pool against your own targets before committing to a plan. Success rates published by a provider are measured on the provider’s targets.
2. Read what the provider allows. Some forbid particular targets outright, and a plan cancelled mid-project costs more than the subscription.
3. Track usage as you go. Traffic plans run out quietly, and a stalled crawl at 90% completion is worse than one that never started.
4. Spread requests over the pool rather than hammering one exit. Rotation only helps if the requests are actually distributed.
5. Keep the proxy layer swappable in code. Providers change hands, rename themselves and occasionally disappear, and a scraper that hard-codes one endpoint has to be rewritten when that happens.

The last one is not theoretical. Two of the services in this market are no longer reachable under their own domains, and a third now trades under a different name.

Conclusion

Pick on pool type, rotation control and billing unit, in that order. Residential addresses are the ones sites treat as ordinary visitors, rotation control decides whether a session survives a multi-step flow, and the billing unit decides who pays for the requests that fail.

Running your own proxy layer means paying for the proxies, for a machine to run the scraper on, and for whatever handles the pages that come back as challenges. A scraping API folds those three into one bill, which is the trade to weigh before you buy traffic by the gigabyte.