Nodetonet for web scraping — rotating mobile proxies that beat datacenter anti-bot

TL;DR: Anti-bot systems block datacenter ASNs on sight. Mobile-carrier IPs share each address among thousands of real subscribers, so vendors won't touch them — making mobile proxies the highest-trust exit type for scraping.

Scraping at any meaningful volume has one core problem: anti-bot systems block datacenter IPs on sight. Cloudflare, DataDome, PerimeterX, Kasada, Akamai Bot Manager — they all maintain ASN blocklists, and AWS / Hetzner / OVH / Linode subnets are at the top. You can pay a residential proxy provider a high per-gigabyte rate to dodge that, or you can route through real mobile-carrier IPs from a phone sitting on Vodafone, Türk Telekom, T-Mobile or Verizon. That is what Nodetonet does.

This page walks through the actual setup: how to pair phones, how to chain a residential upstream for extra capacity, how to design sticky sessions per scrape job, and how to keep the cost low with prepaid credit.

1 · Why datacenter proxies fail at scale

The IP you scrape from determines whether your request gets a 200 or a CAPTCHA. Datacenter IPs are recognised by ASN — anti-bot rules match them in milliseconds. Residential proxies are better but expensive per gigabyte, and increasingly fingerprinted as proxy exits because large pools get recycled and flagged over time.

Mobile-carrier IPs sit in a different category. Carrier-grade NAT (CGNAT) means thousands of real users share each public IP at any moment. Blocking that IP would block paying customers mid-session — anti-bot vendors are extremely cautious about banning carrier ranges. The practical result: mobile IPs see significantly fewer blocks than datacenter IPs on the same protected target.

2 · Proxy type comparison for scraping

3 · How Nodetonet fits the scraping stack

Nodetonet gives you three primitives that compose into a scraping backend:

• Paired devices — your own Android phones running the agent. Each phone is one exit IP that can rotate on demand via airplane-mode cycling. Add more phones for linear throughput scaling.
• Token groups — bundle multiple phones under one proxy endpoint. Requests load-balance across the pool using round-robin or least-connection, with optional sticky sessions when you need IP consistency across a session-aware flow.
• Upstream forwarding — if you need more geographic diversity than your phones provide, chain Bright Data, Smartproxy, IPRoyal or Oxylabs as an upstream residential provider. Nodetonet becomes your single stable HTTPS/SOCKS5 endpoint and manages the sticky-session modifiers on the upstream username automatically.

Rotation mechanics are explained in depth in when and why to rotate tokens and when to use rotating mobile proxies.

4 · Rotating vs sticky — which mode for which job

Choosing the wrong session mode is the most common scraping mistake. Here is the decision tree:

• Rotating (default) — each request goes out on a fresh IP. Use this for crawling product catalogues, SERP scraping, public data collection where each request is independent. Configure with token groups for automatic pool rotation.
• Sticky — one IP pinned to a named session for a set TTL. Use this for login flows, multi-step checkouts, scraping behind auth, and anything where changing IP mid-session triggers a re-auth or CAPTCHA challenge. Switch to sticky just by appending -session-XXXX to your proxy username — no endpoint change required.

Most real scraping workflows mix both: rotate for discovery passes, go sticky for the extraction steps that require authentication.

5 · Step-by-step setup

1. Sign up at /auth/register and add prepaid credit. There is no monthly subscription — idle proxies cost nothing.
2. Install the agent on 2–4 Android phones (/download). Keep them on carrier data for traffic; Wi-Fi is used only for the control channel.
3. Create a token group on the rotating proxies panel, add all phones, and set the rotation interval to match your target's session tolerance (30 s–5 min is common).
4. Create an HTTP or SOCKS5 proxy bound to that token group. Set sticky session to off for pure rotation, or use -session-jobid in the username for sticky-per-job.
5. Optionally add IP allowlists, per-client quotas and thread limits to fence the scraper from the rest of your proxy infrastructure.
6. Point your scraper at the proxy endpoint. Use Proxy Checker to verify the exit IP is a mobile carrier before running production jobs.

6 · Code example — Python with sticky sessions

A minimal Python scraper using the Nodetonet HTTPS proxy with one sticky session per scrape job, so each job's IP stays consistent across its requests:

import requests
import uuid

PROXY_HOST = "your-proxy.nodetonet.com"
PROXY_PORT = 8443
PROXY_USER = "your-token-id"
PROXY_PASS = "your-token-secret"

def scrape(url, job_id=None):
    # Sticky session per job: same job_id = same exit IP for the entire run
    sid = job_id or uuid.uuid4().hex
    proxy_url = (
        f"http://{PROXY_USER}-session-{sid}:{PROXY_PASS}"
        f"@{PROXY_HOST}:{PROXY_PORT}"
    )
    proxies = {"http": proxy_url, "https": proxy_url}
    r = requests.get(url, proxies=proxies, timeout=30)
    return r.status_code, r.text

# 50 pages, each with its own session; IP rotates between jobs
for i in range(50):
    code, body = scrape(
        f"https://target.example.com/p/{i}",
        job_id=f"job-{i}"
    )
    print(i, code)

For Playwright or Puppeteer, pass the proxy at launch: { proxy: { server: "https://your-proxy.nodetonet.com:8443", username: "...", password: "..." } }. Use SOCKS5 if you need full TCP transparency for WebSocket targets or DNS-over-proxy behaviour.

If you are chaining Smartproxy or Bright Data as upstream, configure that in the proxy detail page under Upstream > Provider > Credentials. Nodetonet handles sticky-session modifier injection on the upstream username transparently. See upstream residential forwarding and upstream username modifiers for the full mechanics.

7 · Geo and carrier targeting

Mobile proxies carry real location data — you can request a specific country, city or carrier directly in the proxy username without changing endpoints. Nodetonet supports modifiers like -country-tr, -city-istanbul or carrier-specific pools at Turkcell, Vodafone and Türk Telekom. For the full modifier syntax see geo-targeting.

This matters for scraping localised content: SERP results, pricing, product availability and regional catalogue data all depend on where the request appears to originate.

8 · Advanced controls for production scraping

• Per-client auth and quota — if you run scraping for multiple customers from one proxy fleet, create a separate proxy client per customer with its own username/password, bandwidth quota and expiry date. See proxy clients and per-customer auth.
• Thread limits — cap concurrent connections per client to avoid overwhelming a single device. Explained in thread limits per client.
• Domain allow/deny lists — restrict which domains each proxy client can reach, useful when you want one proxy pool shared across internal and external scraping jobs. See domain restrictions.
• TCP fingerprint spoofing — Nodetonet edge servers can alter the TCP/IP stack fingerprint of outgoing connections, reducing the signal that TLS JA3/JA4 fingerprinting provides to anti-bot systems. See TCP fingerprint spoofing.
• Automatic failover — if a device goes offline mid-job, token groups automatically shift load to remaining online devices. See automatic failover.

9 · When this is not a fit

• You need exits in dozens of countries simultaneously — mobile device coverage follows where phones are physically located. For broad simultaneous geo coverage, chain a residential upstream provider.
• You need massive parallelism from a single pool — device-limited throughput scales by adding phones. At very high request volumes, combine Nodetonet with an upstream residential pool for the overflow.
• The target explicitly blocks mobile carrier ASNs — uncommon, but a small number of platforms now flag specific carrier ranges. Test with Proxy Checker before committing a large job.
• Adversarial use cases (account takeover, credential stuffing, ToS violations) — not supported; accounts are terminated.

For a full platform overview start at Features, verify any IP with the free What is my IP tool, or register and pair your first phone to run a test scrape today.