Why VPNs Keep Getting Blocked in Iran: SNI, DPI, Whitelists and the National Network
Short answer
VPNs in Iran usually stop working because the filtering system recognises the server's IP address, the domain name in the connection, or the shape of the traffic itself, and then blocks or degrades it. SNI filtering, deep packet inspection and UDP throttling are the main tools. Using several protocols with automatic failover is the most reliable way to stay connected.
Anyone who has tried to keep a VPN working inside Iran, or helped family members there do it, knows the pattern. A config works perfectly for a week, then suddenly will not connect. Or it connects, but nothing loads. Or speeds collapse in the evening and recover by morning. None of this is random. Behind almost every failure is one of a handful of well-understood filtering techniques.
This article explains, without assuming a networking background, how Iran's filtering identifies VPN traffic, why some protocols survive longer than others, and what users can realistically do about it.
The big picture: a few chokepoints
Traffic leaving Iran for the global internet passes through a small number of international gateways. That concentration is what makes large-scale filtering possible: every connection can be examined at a few points, and blocked there if needed.
To decide whether to let a connection through, the filtering system looks for clues:
- Destination: the IP address of the server you are connecting to.
- Domain name: the hostname announced at the start of an encrypted connection (the SNI).
- Traffic shape: packet sizes, timing, and the fingerprint of the handshake.
- Transport: whether the traffic is TCP or UDP.
A VPN that looks suspicious on any of these dimensions is at risk.
SNI filtering: when the domain name gives you away
When a browser or app opens an HTTPS connection, it usually announces the name of the site it wants, in a field called Server Name Indication, before encryption is fully established. In most connections that name is still sent in the clear. The filter can read it and drop the connection if the name is on a blocklist.
This is why a VPN provider's website can be unreachable while its servers keep working, and vice versa. The domain has been identified; the server IPs have not, or the other way around.
How REALITY hides behind real websites
VLESS with the REALITY layer was designed for exactly this problem. The connection begins with a genuine TLS handshake that presents the identity of a large, well-known website. To an observer it looks like an ordinary visit to that site, and anyone probing the server without the right key simply sees the real site. Our plain-language guide to VPN protocols goes into more detail.
Deep packet inspection: judging traffic by its behaviour
Deep packet inspection (DPI) means the filter does not stop at the destination; it studies the traffic itself. Even when the content is encrypted, a lot remains visible:
- the structure of the first packets, which acts as a fingerprint for each protocol;
- packet sizes and the gaps between them;
- long, uniform connections that do not resemble normal browsing;
- the ratio of upload to download.
That is why classic VPN protocols such as OpenVPN and WireGuard are usually detected quickly in Iran. They are secure, but they were built for privacy, not for camouflage, and their traffic does not look like anything else on the network.
"Connected, but nothing loads"
A common DPI tactic is not to block a connection outright but to sabotage it: drop packets, throttle speed to a crawl, or reset the connection after a few seconds. It is cheaper for the filter than a hard block and harder for users to diagnose, which is exactly why so many people see a "connected" status with no working internet.
IP blocking: when the server itself is burned
The bluntest method is blocking the server's IP address. Once a server is used by many people, or its traffic pattern draws attention, its address can end up on a blocklist, and then no protocol on that server will work, because packets never reach it. Sometimes whole address ranges belonging to a hosting provider are restricted.
Services that rely on one or two fixed servers therefore do not last long in Iran. A serious provider needs to retire blocked servers and bring in replacements quickly, ideally without the user doing anything.
UDP throttling and mobile networks
Many of today's fastest protocols, including Hysteria2, TUIC and WireGuard, run over UDP, which is excellent on unstable mobile connections. During periods of tightened filtering, however, UDP traffic to foreign destinations is sometimes heavily throttled.
The result is that a UDP-based protocol can work beautifully on one carrier, or at one time of day, and fail completely a few hours later. Behaviour also differs between mobile operators and fixed-line connections, so a config that works on home Wi-Fi may not work on mobile data.
Whitelisting and the "national internet"
Normally, filtering works as a blocklist: everything is allowed except what has been banned. In a crisis the logic can flip to a whitelist, where only approved destinations and protocols get through and everything else is blocked.
In that mode the National Information Network, often called the "national internet", keeps running: domestic messaging apps, banking and local services stay up, while the route to the global internet narrows sharply or closes entirely. The stricter the whitelist, the slimmer the chances for any VPN. During a full international shutdown no VPN can work, because there is no path out of the country at all.
Not every outage is censorship
Before concluding that a server is blocked, check the boring causes:
- Device clock. Several protocols, VMess in particular, refuse to connect if the device clock is off by more than a minute or two.
- Expired subscription or used-up data.
- An outdated client app that does not understand newer settings.
- Too many devices connected at once for your plan.
- Domestic internet problems. Try loading a local Iranian site.
What users can actually do
You cannot make blocking disappear, but you can make outages rarer and shorter:
- Do not rely on a single protocol. Mix TCP-based options (VLESS + REALITY, Trojan) with UDP-based ones (Hysteria2), so that when one is detected another may still pass.
- Prefer apps with automatic failover. Clients that test the network and switch protocols and servers on their own are far more resilient in practice. Zuno's Android and Windows apps do this in Automatic mode, moving to a healthy route when one is blocked.
- Use a subscription link, not a single config. A subscription updates the server list for you; refresh it when things break.
- Try a different network if one carrier is failing.
- Keep a kill switch on so that traffic does not leak unprotected during constant reconnects.
- Check for leaks with a leak test once you are connected.
No provider can honestly promise it will never be blocked. The real difference between services is how quickly, and how quietly, you are moved to a working route after one gets blocked.
If you are helping someone choose a service, our guide on how to choose a VPN for Iran lays out the criteria that matter.
Frequently asked questions
Why does my VPN say "connected" when nothing loads?
Usually because the filter is not blocking the connection outright but dropping or throttling its packets. A wrong device clock or a DNS problem can produce the same symptom.
Why does a VPN work on Wi-Fi but not on mobile data?
Different operators and connection types can face different restrictions, especially on UDP traffic. A config that passes on one network may fail on another.
Is there a protocol that can never be detected?
No. Designs such as VLESS with REALITY make detection much harder, but nothing is permanently invisible. Reliability comes from combining several protocols with fast failover between servers.
Does any VPN work during a full internet shutdown?
No. If international connectivity is cut, there is no route to servers outside the country, so no VPN can connect. Only domestic services remain reachable.