Datacenter Proxies for fast, low-cost lookups on lightly blocked sites

Datacenter IPs come from lines in facilities full of servers, not from homes or phones. Those lines are wide and stable, so responses come back fast and speed holds up even with many requests at once. With no home line or carrier network in the path, the price is the lowest of the four products.

The flip side is that target sites can tell the address comes from a datacenter. Sites that only want ordinary visitors may treat datacenter IPs more strictly. So whether datacenter proxies are the most efficient choice or the wrong one depends on how hard the target blocks.

The connection is the same as the other products: connect to the gate.proxymay.com:7000 gateway over HTTP or SOCKS5 and use your access token as the password. You only change your crawler's proxy setting, so it is easy to compare products with the same code.

The surest way is to measure it with a small sample. Request about 200 target pages over datacenter IPs and record the results.

• If most come back 200 with normal content, datacenter is enough.
• If some 403 and 429 responses mix in, slow down and measure again.
• If you keep landing on login or verification pages, moving to rotating residential saves both money and time.

Targets that usually run well on datacenter IPs include sites you operate, public records from government bodies, pages published for machines to read like product feeds, and sitemaps that allow search engine crawling. Keep your sample results with a date. Blocking rules on the same site can change months later, and re-measuring on a schedule tells you when it is time to switch products.

Datacenter is the product you choose on speed and price. If the job does not need to look like a home user, there is no reason to pay for a pricier address.

If the target often blocks datacenter addresses, switch to rotating residential proxies: each request leaves from a different home IP and blocks drop. If you need the same IP logged in for weeks, use static ISP proxies; if you need to see what phone users see, use mobile proxies.

In practice, people often mix them. Fetch the light parts, like sitemaps and listing pages, quickly over datacenter IPs, and send only the detail pages that get blocked to rotating residential. Total cost drops a lot, and with one balance there is no separate top-up for each.

• Checking your own site: confirm on a schedule that your site loads in several countries and how fast it responds.
• Sitemap crawling: walk sitemaps with tens of thousands of URLs quickly to find broken links and redirects.
• Public listing collection: pull large volumes of data that the publisher allows to be collected, like public records and open product feeds.
• Localization testing: check that language, currency and price display switch correctly by country on your own service.

Datacenter proxies shine when you run many parallel requests, but the parallel count should stay within what the target can handle. Here is a Scrapy example that spreads requests across several IPs while controlling speed.

# settings.py
CONCURRENT_REQUESTS = 32
CONCURRENT_REQUESTS_PER_DOMAIN = 8
DOWNLOAD_DELAY = 0.25
AUTOTHROTTLE_ENABLED = True
RETRY_HTTP_CODES = [429, 503]

# proxies.txt (copied from the dashboard in host:port:user:token format)
gate.proxymay.com:7000:USERNAME:YOUR_TOKEN

With several IPs, split requests evenly between them. If one IP takes most of the load, that address hits limits first.

Datacenter proxies are billed per IP, from $1.20 per IP. Data transferred is not part of the price, so the bigger the pages and the more requests you make, the more it beats per-GB products.

For example, buy 20 IPs to run a sitemap crawler and the cost works out to 20 × $1.20. Pull tens of GB through those IPs and the amount stays the same. The same volume on rotating residential, which is billed per GB, would keep drawing down the balance for every GB.

So with lots of data and light blocking, datacenter usually wins; with little data or heavy blocking, rotating residential often costs less. A sample run that measures block rate and page size lets you decide with numbers.

1. 1Sign up and store your access token. It is both your login and your proxy password.
2. 2Top up, then choose a country and the number of IPs in the datacenter product.
3. 3Copy the dashboard format your tool needs: http, socks5 or host:port:user:token.
4. 4Add it to your crawler or monitoring tool and run the curl check to confirm the IP location.
5. 5Start with few parallel requests and raise them step by step while watching response codes.

Start the parallel count low and raise it only after the error rate looks steady. Where to enter settings in each tool is covered in Quick start.

Datacenter proxies are available in 18 countries, centered on places with many server facilities such as the United States, the United Kingdom, Germany, Japan and Singapore.

Choose on two grounds. If you need what users in a country see, as in localization testing, pick that country. If raw speed matters, pick a country close to the target's servers for faster responses. Products per country are listed on Coverage, and a page like Singapore IP shows at a glance which products you can use there.

Because datacenter proxies are fast, they can also put load on a target easily. Finding the rate at which the target still answers normally gets you more data in the end than running at maximum speed.

• When 429 responses rise, halve the parallel requests, then raise them again slowly.
• Set a separate cap on requests per domain.
• Respect robots.txt, the terms of service and what is meant to be public.
• Load testing or attacks that could take a target down are forbidden.

If blocks keep coming, do not force it. Slow down or move to another product.

Run a sample job on a few IPs and measure the block rate first. If the results are good, add IPs to scale up and send only the blocked parts to another product. See top-up sizes on Plans, then sign up to get your token.