VPN & blocking 9 min read By KRAKOBIT

Mobile internet whitelists: how Russia and Iran build them

Mobile internet whitelists in Russia and Iran: who draws up the lists, how filters decide what passes and how the two models differ.

A mobile internet whitelist flips the usual logic of censorship: everything is forbidden except what has been approved in advance. In Russia this mode is introduced region by region, and the authorities justify it with the threat of drone attacks. The phone still shows 4G, Russian services from an approved list open, while foreign websites and VPNs to servers abroad fail to connect. In Iran the whitelist is built into the network architecture itself: all external traffic passes through a single state gateway, and during major shutdowns the country is left with domestic services only. Below: how such lists work technically, who draws them up and how the Russian model differs from the Iranian one.

How is a whitelist different from ordinary blocking?

Ordinary blocking runs on a blacklist. The censor keeps a register of what is banned: websites, server addresses, recognisable VPN fingerprints. Everything else is allowed. A new entry appears only after a site or service has become noticeable, so the censor is always catching up.

A whitelist works the other way round. A short list of services is allowed, everything else is dropped by default. The censor no longer has to guess whether traffic looks like a VPN. It only has to check whether the destination is on the list.

Blacklist Whitelist
Default everything allowed everything blocked
New site or server works until noticed fails until added
Traffic disguise helps nearly useless
List size very large, keeps growing short, can be checked by hand
Cost to the economy low high: anything relying on foreign infrastructure breaks

The last row explains why whitelists are not switched on everywhere and forever. An internet with a limited set of open services is expensive. Foreign cloud software for business, app updates and services that depend on someone else’s servers all stop working. So both Russia and Iran treat the whitelist as a mode that is switched on and off, not as a permanent state of the network.

How does a filter decide what to let through?

The word ‘list’ hides three different checks. In practice they are combined.

Where is the traffic going?

Every server on the internet has a numeric address, an IP. Large companies own whole blocks of these addresses. The crudest and most reliable method is to allow traffic only to blocks owned by approved companies and clouds. Data heading to any other address never gets past the operator. Using someone else’s name does not fool this check: only the address matters.

What is the site called?

When a phone opens an encrypted connection to a website, its very first message openly states the site’s name, for example ozon.ru. Deep packet inspection equipment, DPI, reads that name and compares it with the list. This check is more precise than the address check, but a false name can be supplied. That is why the two are combined.

What does the traffic look like?

The third check looks at the shape of the data: does the start of the connection resemble opening a normal website, an address lookup or a standard encrypted connection. Anything that looks like nothing familiar is cut. For such a filter, traffic disguised as random noise is suspicious in itself.

Why does every list leak?

A modern app rarely lives at one address. A banking app gets notifications through Google or Apple servers, maps pull images from content delivery networks (CDNs), captchas and analytics go to third-party companies. If even one dependency is missing from the list, the app half-loads or hangs.

The flip side: big CDNs serve thousands of customers from the same addresses. Allowing a CDN for one service opens the door for other sites on the same infrastructure. So whoever compiles the list is always choosing between two mistakes: letting too much through or breaking what was meant to work.

How does Russia build mobile internet whitelists?

The infrastructure: TSPU

The technical basis arrived with the ‘sovereign internet’ law, which took effect on 1 November 2019. Under it, telecom operators host TSPU boxes, the ‘technical means of countering threats’. This is DPI equipment centrally controlled by Roskomnadzor rather than by the operator.

When What happened
1 November 2019 ‘sovereign internet’ law takes effect, operators install TSPU
March 2021 Twitter throttled, the first high-profile test of TSPU
2024 YouTube throttled

Throttling a single service and running a whitelist are technically related. In both cases equipment inside the operator’s network looks at where traffic is going and decides what to do with it. The difference is that throttling has one target, while a whitelist inverts the rule: only what is listed is allowed.

Who draws up the list and who switches the mode on?

The Russian model has two roles. The Ministry of Digital Development compiles a list of socially significant services that must keep working while mobile internet is restricted. Decisions on where and when to impose restrictions are taken by regional authorities, usually on security grounds.

The operator is the executor in this scheme. If filtering is managed centrally, the mode does not depend on whose SIM card is in the phone.

The exact composition of the list is not given here: it changes, and the current version is published by the ministry itself. The selection principle follows from the technical logic: only what is hosted inside Russian networks can work in this mode.

What does a service need to get on the list?

Being on the list does not make a service work by itself. Recall the section on leaks: if an app pulls notifications, images or analytics from abroad, it will break even with permission. So a service needs servers inside the country and as few external dependencies as possible.

Hence the regulator’s interest in data centres. KRAKOBIT has already reported that Russian data centres may be required to reserve 10% of capacity for whitelists. If the mode is switched on regularly, approved services need guaranteed capacity inside the country.

The data centres themselves sit in the same risk zone used to justify the connectivity restrictions. Russia is discussing how to move data centres underground to escape Ukrainian drones.

What does the mode look like to a subscriber?

From the outside a whitelist looks like an outage: there is signal, but most apps will not load. A similar picture came from the recent NTT Docomo outage that disrupted calls, SMS and mobile data in Japan. The difference is fundamental. In an outage the network is broken. Under a whitelist the network is healthy and works exactly as configured.

Officially, Russia speaks of restrictions on mobile internet specifically. There are two scenarios: a complete cut of mobile data, or a mode in which services from the list keep working.

How does Iran build its internet whitelist?

One door to the outside

Iran built its system differently. The country’s international connectivity passes through the state-owned Telecommunication Infrastructure Company, TIC. It is one door to the outside, and the state holds the key. Inside the country runs the National Information Network: domestic data centres, the messengers Bale, Eitaa, Rubika and Soroush, the ride-hailing app Snapp, the Digikala store, banking services. Policy is set by the Supreme Council of Cyberspace.

This architecture lets the country be cut off from the outside world without breaking domestic life. From outside it looks like an internet shutdown. From inside it looks like an internet with only Iranian services left. In effect it is a whitelist where the entire domestic network plays the role of the list.

Timeline of major shutdowns

When What happened
November 2019 fuel price protests, international internet cut for almost a week, domestic services keep running
September 2022 protests after the death of Mahsa Amini: mobile internet shutdowns, Instagram and WhatsApp blocked
December 2024 WhatsApp and Google Play unblocked
June 2025 12-day war with Israel: Iran’s external traffic, according to NetBlocks and Cloudflare, drops almost to zero

The 2019 shutdown was the first full-scale test of the National Information Network. Since then Iran has repeated the scenario at different depths: from evening mobile shutdowns to an almost total loss of external traffic.

Filtering by traffic type

In 2020, researchers from the University of Maryland’s Geneva project described an Iranian ‘protocol whitelist’. On popular ports the filter lets a connection through only if its start resembles one of the familiar traffic types: a site address lookup, ordinary web or a standard encrypted connection. Anything unfamiliar is dropped.

This is the third check from the section above in its pure form. Iran does not list allowed sites here, it lists allowed kinds of traffic.

A whitelist of people

Another Iranian idea: different internet for different people. Inside Iran it is called ‘class-based’ or ‘tiered’ internet. Universities, companies and certain professions can get access with less filtering.

In November 2025 an X feature showing an account’s country revealed that some pro-government users and officials were accessing X directly from Iran, even though X itself is blocked in the country. On Iranian social media this was attributed to ‘white SIM cards’: SIMs for which filtering is switched off. Here the whitelist is made not of addresses but of people.

Russia and Iran: what is the difference?

Russia Iran
Infrastructure TSPU at operators, controlled by Roskomnadzor single TIC gateway and the National Information Network
Who compiles the list Ministry of Digital Development Supreme Council of Cyberspace and relevant agencies
Who switches the mode on regional authorities central government
Official reason threat of drone attacks protests, war
Scale individual regions the whole country
Which networks mobile internet mobile and fixed during major shutdowns
What keeps working Russian services from the list domestic services: Bale, Eitaa, Rubika, Snapp, Digikala
Access by lists of people not publicly described ‘white SIMs’, access for selected groups
How filtering works service addresses and names addresses, domestic network, traffic type

In short: Russia builds its whitelist from below, through equipment at operators and decisions by regions. Iran builds it from above, through a single exit point from the country. The Russian model is more flexible: it can be switched on in one region and leave the neighbouring one alone. The Iranian one is cruder but has been tested on major shutdowns since 2019.

Why do VPNs stop connecting?

Against a blacklist, a VPN competes on disguise: look like an ordinary website, hide recognisable features, use the name of a popular service. Which protocols get through the filters of China, Russia and Iran in normal mode is covered in the article on censorship emulation.

Under a whitelist, disguise runs into the address. A connection can carry the name of an allowed site, but the data still goes to a server whose address is not in the allowed blocks. The filter checks not only the name but also where the traffic is actually going.

So in this mode what matters is not the protocol but where the server physically sits and whether its network is on the allowed list.

Where are whitelists heading?

In Russia the whitelist is no longer improvised: there is a ministry-approved list, a procedure by which regions impose restrictions, and a discussion of reserving data centre capacity for this mode. Each of these steps makes the mode easier to switch on again.

Iran shows a later stage of the same logic. The domestic network lives without the outside world, and access to the external internet is distributed by user group.

The two models share one thing: the whitelist stops being an emergency measure and becomes a network setting that can be switched on by government decision. For any service that wants to work in such a mode, the key question is no longer its code but whose network its servers sit in.