Bangladesh proxies
Bangladesh breaks two comfortable assumptions. The first: that the digits on a page are the digits your regular expression is looking for. Here they are Bengali, and a search for Arabic numerals finds neither the price nor the date. The second: that a residential bill depends on how many pages you fetch. Here it depends on what they weigh, and the gap between a careful and a careless collector is measured in multiples.
What we have in Bangladesh
Prices are the same as in every other country we cover.
What Bangladeshi addresses get used for
Marketplace price monitoring
Daraz BD, Pickaboo, Chaldal, Rokomari: Bangladeshi storefronts and prices are only visible from inside the country.
Classifieds and second-hand
Bikroy.com holds most private listings: goods, cars and jobs.
Property
Bproperty: rentals and sales in Dhaka, a market that turns over quickly.
Local search and ad verification
google.com.bd, Prothom Alo, The Daily Star, Bdnews24: campaigns are bought for the local audience.
What looks different from outside Bangladesh
These serve a Bangladeshi visitor something other than what a foreign one sees, because of broadcast rights and local catalogues. We sell addresses rather than a way past detection.
Which type to take for a Bangladeshi job
Home connection addresses, billed per gigabyte, rotating by default. A sticky session holds one address for up to fifty minutes.
Marketplaces, classifieds, search results, ad checks
Addresses in Bangladeshi datacenters, traffic unmetered.
Volume collection where the target ignores address class
Cities people ask for here
Cities are set with the _city_ suffix. Dhaka outweighs everything, Chattogram adds a noticeable amount, and beyond them the pool is thin.
The setups people take here most often
- 01Residential with images blockedThe default, and blocking media here is a necessity rather than an optimisation.
- 02Datacenter for volumeWhere the target is neutral: traffic is unmetered and page weight stops mattering.
- 03A fifty minute sticky sessionWhen a job has to run start to finish on one address.
- 04Digit normalisation on intakeBengali numerals are converted before parsing rather than after.
Connection strings for Bangladesh
The country code is BD, upper case. The city goes after it with _city_, hyphens instead of spaces, in the order country, city, session, lifetime.
Bangladeshi static addresses take no suffixes: the host is the IP from your dashboard, port 50100, SOCKS5 on 50101.
Bangladeshi exits, four ways
curl -x login_c_BD:[email protected]:10000 https://api.ipify.org
# held for fifty minutes, for a job that must not change address
curl -x login_c_BD_s_42_ttl_50m:[email protected]:10000 https://api.ipify.org
# Chattogram explicitly, when regions are compared
curl -x login_c_BD_city_Chattogram_s_43_ttl_15m:[email protected]:10000 https://api.ipify.org
# a Bangladeshi static address: no suffixes
curl -x login:password@YOUR-BD-IP:50100 https://api.ipify.orgWhat you capture is not what you see
The first feature, and it is about your parser seeing something other than what a person in a browser sees.
Bengali writing has its own digits. A price, a date or a phone number on the page may be written in them, and a regular expression over Arabic numerals finds nothing even though the number is on screen. Worse, both sets usually appear on the same page: the heading in Bengali digits, a technical field in Arabic ones.
The cure is normalisation on intake. Convert digits to one form before parsing rather than after, and use Unicode-aware character classes instead of spelling out zero to nine.
The second half of the problem is older and nastier. Some sites, and documents especially, still use pre-Unicode font encodings: the bytes are Latin, and the text only looks Bengali because of the font. Your scraper captures a run of Latin characters that read as perfectly normal Bengali in the browser. It resembles a broken character set, but the cause is different and ordinary re-encoding does not fix it.
The practical conclusion: check what you captured rather than what was rendered. If the body contains meaningful Bengali words, you are working with Unicode. If it is Latin soup while the browser shows Bengali, flag that source separately and handle it on its own rather than trying to repair it with a general rule.
On dates specifically: some resources print a Bengali calendar year alongside the Gregorian one. Take the date from the machine-readable attribute and range-check whatever you read from visible text.
Page weight decides your bill here
The second feature, and it is about money directly.
Residential is billed by the gigabyte rather than by the request. In countries with light pages the gap between a careful and a careless collector is a few percent. Here pages are often heavy: large uncompressed images, embedded fonts for Bengali text, plenty of third-party scripts. The gap becomes a multiple.
The arithmetic is simple. A page with a megabyte and a half of images against the same page without media at a hundred kilobytes: a fifteenfold difference in the bill for an identical set of data. At twenty thousand listings a day that is thirty gigabytes against two.
So blocking images, fonts, video and ad scripts here is part of the initial setup rather than an optimisation for later. If you work through a headless browser, disable media loading at the request interception level rather than hoping the cache handles it.
And a check before scaling: fetch twenty pages, look at the actual volume and multiply. An estimate built on page count without measuring weight is wrong by more in this country than anywhere else.
If the target ignores address class, the datacenter option removes the question entirely: traffic there is unmetered and page weight stops affecting the bill.
What actually blocks you on Bangladeshi targets
The familiar South Asian split.
Marketplaces and Bikroy filter hosting ranges: expect a 403 from datacenter and plan for residential.
Publishers and government sites serve datacenter addresses normally.
Search results are reachable but rate limited per address: more addresses rather than a different type.
Some platforms only deliver listings after scripts run, which is another reason to watch volume: rendering drags in everything you did not block.
Which exits you get here, and how stable they are
The Bangladeshi pool in practical terms.
The pool is decent in size thanks to the population but geographically narrow: the overwhelming share sits in Dhaka and Chattogram.
Line quality varies and response times fluctuate more than in neighbouring countries. Thirty seconds is the minimum timeout here, and retries belong in the architecture.
Response times from Europe are long: the country is far away and traffic transits other networks.
Sticky sessions hold an address for up to fifty minutes, and default rotation changes it on every request.
Choosing the type for a Bangladeshi target
The usual order, with a financial fork in it.
- Start with a Bangladeshi datacenter address: $1.15 per address per month. On publishers and reference sites that is enough, and page weight does not matter there.
- A 403 on the first request means hosting ranges are filtered. Move to country-level residential.
- On residential, configure media blocking immediately: it is the main lever on the bill here.
- If the job has to run on one address, add a sticky session with a lifetime.
- Measure the weight of twenty pages before scaling: an estimate by page count misleads more here than anywhere.
What Bangladeshi sites check beyond the address
A Bangladeshi address opens the first gate; language, digits and time decide the rest.
Send bn-BD. English versions are widespread here, particularly in business sections, so without an explicit header you will collect the wrong variant.
The currency is the taka, written ৳, and the amount beside it may be in Bengali digits.
The time zone is UTC+6 with no daylight saving, one zone for the whole country.
The weekend here is Friday and Saturday, with Sunday a working day: build your schedule and weekly slices on the local week, or conclusions about business activity shift.
What a month of Bangladeshi work costs
Three scenarios at our prices, and here they are unusually instructive.
- Price monitoring, 20,000 listings a day with images blocked, roughly 100 KB per page. That is about 2 GB a day, around 60 GB a month: $60 on residential.
- The same 20,000 listings without media blocking, roughly 1.5 MB per page: about 30 GB a day and 900 GB a month, meaning $900. The data you get is identical.
- If the targets are neutral, the same collection on ten datacenter addresses costs $13.50 a month regardless of page weight.
The legal part, stated plainly
Bangladesh is not in the European Union, and GDPR does not apply here directly. Collecting data that identifies people in the country is governed by local law regardless of where your company sits.
This is general information rather than legal advice. Our terms require lawful use and we act on abuse reports. Where the answer carries commercial risk, ask a lawyer.
Verifying a Bangladeshi exit
Under a minute. The second command matters most here: it shows the actual weight of a page, which is what your bill depends on.
Three checks
# 1. is the exit Bangladeshi, and where
curl -s -x login_c_BD:[email protected]:10000 https://ipinfo.io/json
# 2. what one page actually weighs through this exit
curl -o /dev/null -s -w "bytes: %{size_download}\n" -x login_c_BD:[email protected]:10000 https://your-target.com.bd/product
# 3. does the fifty minute session hold
for i in 1 2 3; do
curl -s -x login_c_BD_s_42_ttl_50m:[email protected]:10000 https://api.ipify.org; echo
doneQuestions people ask before buying Bangladeshi addresses
Why does my regex not find the price?
The price is most likely written in Bengali digits. Normalise digits on intake and use Unicode-aware character classes.
Why does the text look like Latin soup?
Some sources use pre-Unicode font encodings: the bytes are Latin and the font supplies the Bengali appearance. Re-encoding does not help; flag those sources separately.
Why is my bill higher than expected?
You pay by the gigabyte here and the pages are heavy. Without media blocking the same dataset costs several times more.
How do I estimate the budget before starting?
Fetch twenty pages, measure the real volume and multiply. An estimate by page count is more misleading in this country than anywhere else.
How should I schedule collection?
On the local week: the weekend is Friday and Saturday, and Sunday is a working day.
Do Bangladeshi addresses cost more?
No. Prices are the same everywhere: residential from $1.00 per GB, datacenter from $1.15 per address per month.