FAQ

Proxy questions, answered

Proxies carry a lot of jargon, and most explanations either oversimplify or bury you in detail that does not help you decide anything. These are the questions people actually ask before buying or setting one up, answered against how the NullVault proxy network actually works.

What are residential proxies?

A residential proxy routes your requests through an IP address tied to a real home internet connection rather than a data centre. Instead of your traffic arriving as an obvious server, it arrives looking like a household in whatever country or city you are targeting, which is why it gets blocked far less often than a datacenter address and why it suits location specific content.

You can run them rotating, meaning a new address on every request, or sticky, meaning the same address for a whole session. The choice is not about quality, it is about whether consecutive requests should look like one visitor or many. Market research, ad verification, search monitoring and collecting public data at scale are the usual jobs.

What to ask a provider

How the addresses are sourced, and whether the people whose connections carry your traffic agreed to it. That question separates networks quickly, and a provider that will not answer it plainly has told you the answer.

Worth knowing

Residential proxies do not make anything unlawful acceptable and they are not an anonymity switch. They are a networking tool, not a shortcut around the law, and any provider implying otherwise is selling you a problem.

How do I scrape websites without getting blocked?

Work with the site rather than trying to overpower it. Check the terms and robots.txt, know what is lawful before you pull anything, slow your requests, keep concurrency modest, cache what you already have, and stop re downloading pages that have not changed.

A proxy helps with geographic testing and with spreading legitimate traffic across more addresses. It does not fix careless request behaviour, and no pool is large enough to outrun a pattern that looks automated.

Four habits make more difference than most people expect.

Keep sessions stable for anything multi step, rather than changing identity partway through a flow.
Back off on failure in measured steps instead of retrying harder, which is how a slow down becomes a block.
Watch your response codes rather than only your throughput. A page quietly returning cached or default content is more dangerous than one returning an error, because it poisons your data without alerting you.
Collect only what you need, and respect the rules covering personal data.

Worth knowing

If the site publishes an official API or a data export, use it. It is faster, more reliable, and it does not depend on the site never changing its markup.

Can I use proxies anonymously without KYC?

There is no version of anonymous that is guaranteed, and skipping identity checks is not automatically privacy done properly. Know your customer checks are how many providers handle fraud prevention and payment compliance. Rather than hunting for whoever asks the fewest questions, look for a published policy on how long data is kept, who owns the company, and how abuse is handled.

A few things hold regardless of who you buy from.

Use HTTPS to the destination. A proxy carries your traffic, it does not encrypt it for you.
Protect your credentials, and never put them in a URL that lands in shell history or a log.
Do not send sensitive information through third party infrastructure.
Remember that websites, payment processors and network operators keep their own records even when your provider asks you for nothing.

Worth knowing

Proxies are not a tool for concealing fraud, evading account bans or avoiding law enforcement, and no KYC does not change that. If what you want is everyday personal privacy rather than a business capability, a reputable VPN, encrypted connections and careful account habits will serve you better than any proxy.

NullVault or Decodo, which is better for scraping?

We would say NullVault, and you are reading that on NullVault’s own site, so weigh it accordingly. On the figures that are checkable, we are cheaper per gigabyte, we do not require ID, and we offer rotating and sticky sessions with country targeting across residential, ISP and mobile pools.

Decodo, which is what Smartproxy is now called, has a wide feature set and a great deal of published documentation, and some teams value that highly. Documentation does not make the underlying addresses perform better, but it does make a team faster, and that is a real advantage rather than a nothing.

Where they beat us

They take card payments and issue conventional invoices, which we do not. If you need a receipt for an expense report or your finance team will not process crypto, that decides it regardless of price per gigabyte. They also have years of independent reviews behind them and we do not.

The test that actually settles it

Run both against the same URLs and compare successful response rate, latency, how often you see a CAPTCHA, and total cost for the same work. That measurement beats anybody’s opinion including ours, it takes an afternoon, and both of us sell small enough quantities to make it cheap.

Worth knowing

This is a comparison written by one of the two providers being compared. Treat it as our argument, not as an audit, and run the test above before committing to volume.

Pay as you go or subscription proxies, which should I pick?

Pay as you go charges for what you use, usually by bandwidth, and suits occasional, seasonal or unpredictable work because you are not paying for capacity that sits idle. A subscription charges a recurring fee for a fixed allowance and tends to win once usage is steady and high, sometimes with better support or higher limits attached.

A subscription is not automatically cheaper, and it is most often the wrong choice for somebody using a fraction of what they pay for. Estimate a month of real usage first, then run both sets of numbers against that figure rather than against the plan you hope to grow into.

Compare these before deciding, because the headline rate hides most of them.

The real price per gigabyte at the volume you will actually buy, not at the top tier.
The minimum purchase.
Whether unused bandwidth expires or carries over.
Concurrency limits.
Location coverage for the countries you need rather than the total count.
Refund terms.

Worth knowing

Per gigabyte plans where credits never expire remove the main disadvantage of the model, which is the pressure to use bandwidth before it disappears. Ours work that way, which is a point in our favour and worth verifying rather than taking from us.

What are IPv6 proxies, and when do I need them?

IPv6 proxies route traffic through IPv6 addresses rather than the older IPv4 ones. The address space is vastly larger, so allocations cost the network very little, which is why they are the cheapest addresses available and why you can have a great many distinct ones at once.

They fit large scale testing, crawling services that already speak IPv6, and anything needing many separate addresses cheaply. They work when the target site, the DNS setup, the hosting and your own software all handle IPv6 properly, and that is four things rather than one.

The check that decides it

IPv6 support is still not universal. Some sites treat IPv6 traffic differently or serve it incomplete content, and many do not answer over it at all. Confirm your scraper, proxy library, TLS settings and DNS resolver handle it cleanly, then confirm your actual targets do, before buying anything.

Worth knowing

Use IPv6 because the job calls for it, not because the pool is bigger. A cheaper address that cannot reach your target costs more than an expensive one that can.

How do I integrate a proxy into a Python scraper?

Take the provider’s endpoint, your credentials, the supported protocol and any limits. Keep the credentials in environment variables rather than in the script, then pass the proxy through a dictionary, set a timeout, and decide up front which status codes you will accept.

Then the parts that decide whether it survives contact with a real site.

Keep concurrency modest and retry with backoff rather than immediately trying again.
Log the status, the latency and which session or location you used. Never log the credentials.
Validate what comes back before storing it, because a proxy returning a block page still returns a 200 sometimes.

python

import os
import requests

# From the environment, never hardcoded. A proxy URL contains a password,
# and a hardcoded one ends up in version control within the week.
proxy_url = os.environ["PROXY_URL"]  # http://user:pass@gateway:port

proxies = {
    "http": proxy_url,
    "https": proxy_url,
}

# A timeout is not optional. Without one a hung connection blocks the worker
# for as long as the operating system allows, which can be minutes.
response = requests.get(
    "https://example.com",
    proxies=proxies,
    timeout=15,
)

The https key selects the proxy used for https destinations. It does not mean the hop to the proxy is encrypted, which is the single most common misreading of this dictionary.

Worth knowing

If the site offers an official API, use it instead. Every scraper is a maintenance commitment that the site can break without warning.

What is the right proxy setup for price monitoring?

Reliability, geographic accuracy and compliance matter more than pool size. Datacenter addresses are fine for public pages with light protection, and they are a fraction of the cost. Move to residential or ISP where regional pricing or stronger defences are in play, rather than starting there.

Match the address location to the market you are tracking, because a price fetched from the wrong country is not a slightly worse answer, it is a different number. Use sticky sessions where a sequence of requests should look like one visitor, such as anything involving a cart.

What to record

Price, currency, stock status, shipping assumptions, the timestamp and whether the request succeeded. Without the last two you cannot tell a price change from a collection failure later, and that distinction is the whole value of the dataset.

Before committing to a larger plan, run a small pilot and measure.

Successful page loads as a share of attempts.
Whether the regional pricing you got back is actually correct for that market.
Latency and bandwidth cost per page.
How often you hit a CAPTCHA.
How complete the extracted data is.

Worth knowing

Schedule checks sensibly, cache what you already pulled, respect rate limits, and collect only what is publicly available. Retried requests are billed like any other, so a crawl fighting a site can spend a surprising share of its budget on attempts that returned nothing.