Skip to content
AdminKit

TR-069/CWMP (port 7547): why ISPs use it, and when it's a risk

AdminKit explains TR-069, also called CWMP, the protocol many ISPs use over port 7547 to remotely configure the router they gave you — genuinely useful for the ISP, but in November 2016 a flaw in how some routers handled TR-069's related TR-064 commands let a Mirai botnet variant knock roughly 900,000 Deutsche Telekom customers offline in a single weekend.

By AdminKit team, Netbase JSC · Updated

Key takeaways

  • TR-069 (CWMP) is how many ISPs manage the routers and modems they supply. The router calls the ISP’s Auto-Configuration Server (ACS); the ISP can ask it to call in sooner by sending a “Connection Request” to the router.
  • That Connection Request listener is why port 7547 is often open on the WAN side of ISP routers. IANA assigns 7547 to CWMP, and the specification lets the router use it.
  • The protocol requires authentication, but a reachable listener is still code that anyone can send packets to. In 2016 a Mirai variant abused port 7547 on vulnerable routers and knocked about 900,000 Deutsche Telekom lines offline.
  • You usually cannot turn TR-069 off on an ISP router. Check whether 7547 answers from the internet, then ask your ISP to restrict it, or use your own router behind the ISP device.

What is TR-069/CWMP for?

CWMP, the CPE WAN Management Protocol, is almost always referred to by the number of the technical report that defines it: TR-069. It exists to solve a real operational problem for ISPs — once a router or modem ships to a customer’s home, how does the ISP update its firmware, change its configuration, or diagnose a fault without sending a technician or walking a customer through settings over the phone? TR-069 answers that by letting the router periodically contact an Auto-Configuration Server (ACS) run by the ISP, over HTTP or HTTPS, and accept management commands from it during that session. IANA registers TCP and UDP port 7547 as cwmp, and the router side of the protocol commonly listens there, as the next section explains. For straightforward support operations — pushing a firmware patch to every deployed unit of a given model, for instance — it’s a genuinely useful piece of infrastructure, not a design mistake.

Why does a router listen on port 7547 at all?

If the router always starts the conversation, why is anything listening? The answer is in section 3.2 of the Broadband Forum’s TR-069 specification. Sessions are always set up by the router (the “CPE”), which acts as the HTTP client while the ACS acts as the server. But an ISP sometimes needs a change to happen now, not at the next scheduled check-in. For that, the specification defines a Connection Request: the ACS sends a small HTTP request to a URL published by the router, and the router responds by opening a fresh session back to its pre-configured ACS. The specification notes that “Port 7547 has been assigned by IANA for the CPE WAN Management Protocol” and that “the CPE MAY use this port in the Connection Request URL.”

So on many ISP routers, port 7547 on the WAN interface is not the management session itself; it is the doorbell that tells the router to call home. The design has sensible protections written in. The Connection Request carries no commands, and the router only ever calls the ACS address it already knows. The specification also says the router “SHOULD restrict the number of Connection Requests” it accepts in a period, to limit denial of service. And it must check credentials: “The CPE MUST accept Connection Requests from any source that has the correct authentication parameters for the target CPE.”

Read that last requirement carefully. It means the listener accepts connections from any source and then decides based on credentials. Unless the ISP adds a network filter, everyone on the internet can reach the code that parses those requests. That is fine while the code is correct. It is exactly what went wrong in 2016.

MessageDirectionPort / transportWhat it does
Inform (periodic or event)Router → ACSACS URL, HTTP or HTTPSStarts a session; reports status and events
RPCs such as GetParameterValues, SetParameterValues, Download, RebootACS → router, inside the sessionSame sessionReads or changes configuration, pushes firmware
Connection RequestACS → routerRouter’s Connection Request URL, often TCP 7547Asks the router to open a session now
Alternative Connection Request (XMPP, UDP)ACS → router via other pathsDefined in later amendmentsReaches routers behind NAT without an open inbound port

When it became a very public risk

In late November 2016, that same management channel became the entry point for one of the more dramatic ISP-facing incidents of the decade. As reported by WeLiveSecurity, roughly 900,000 Deutsche Telekom customers in Germany lost internet, phone and TV service over a single weekend after a Mirai botnet variant began scanning the internet for devices listening on port 7547 and sending them a crafted command over the related TR-064 protocol — a local configuration protocol that, on certain vulnerable router models, accepted unauthenticated commands capable of injecting and executing arbitrary code. Coverage from The Hacker News at the time confirmed the same mechanism: a modified version of the Mirai worm, previously known for large DDoS attacks built from compromised IoT devices, had pivoted to exploiting a router-management protocol instead of only brute-forcing Telnet logins.

Notably, Deutsche Telekom’s own routers weren’t actually susceptible to the code-injection flaw itself — they crashed from the sheer volume of malicious scanning and exploit-attempt traffic flooding their management interface, becoming collateral damage rather than successfully compromised devices. Other ISPs’ equipment, including Zyxel and Eir (Irish telecom) routers, were directly vulnerable to the underlying flaw. The academic analysis of the Mirai botnet situates this TR-069/TR-064 variant as part of Mirai’s broader evolution beyond its original Telnet-scanning behavior, expanding to whatever internet-reachable management interface a given device family exposed.

What makes port 7547 risky versus merely present

The protocol itself isn’t the vulnerability — a TR-069 connection from a router out to its ISP’s ACS, initiated by the router and restricted to the ISP’s own infrastructure, is standard practice across the industry. The risk comes from two conditions stacking together: port 7547 being reachable from the general internet rather than only the ISP’s management network, combined with a router implementation that handles inbound commands on that port insecurely. Either alone is a narrower problem; both together is what let a single botnet scan and compromise devices at scale in a few days.

Firmware handles 7547 safelyFirmware has a flaw on 7547
7547 filtered to the ISP’s ACSNormal ISP managementRisk limited to whoever can reach the ISP’s management network
7547 reachable from the internetExposed listener; depends on the code staying correctThe 2016 situation: mass scanning finds and hits the flaw

How do I check my own exposure?

  1. Identify your current public IP using AdminKit’s what is my IP tool — this is the address any reachability test needs to target.
  2. Test whether port 7547 responds from outside your network with AdminKit’s port check tool, the same check covered step by step in is port 7547 (TR-069/CWMP) exposed?. A response from the general internet (as opposed to only being reachable from your ISP’s own address ranges, which this type of test can’t fully distinguish) is worth following up on with your ISP directly.
  3. Ask your ISP what’s listening there and who can reach it. If you operate your own router instead of an ISP-provided one, TR-069 typically isn’t running on it at all, which sidesteps the question entirely.
  4. Keep router firmware current. Whether a management port is dangerous if reached depends heavily on whether the firmware behind it has been patched against known flaws like the one exploited in 2016.

What a response on 7547 looks like

If you want more than “open or closed”, send one request with curl from outside your network to your own public address (a documentation address is shown). A Connection Request listener normally refuses unauthenticated requests:

$ curl -s -I --max-time 5 http://203.0.113.5:7547/
HTTP/1.1 401 Unauthorized
WWW-Authenticate: Digest realm="...", nonce="...", qop="auth"
Content-Length: 0
What you seeWhat it meansWhat to do
An authentication challenge (401 with WWW-Authenticate)A CWMP listener is reachable from the internet and asks for credentialsAsk your ISP to filter 7547 to its ACS addresses; keep firmware current
A web page or banner that is not an authentication challengeSomething other than a standard Connection Request handler is answeringIdentify the device; this is worth raising with the ISP or vendor
Connection refused or timed out7547 is not reachable from your test locationGood; re-check after firmware updates or a router swap

Do not try to authenticate or send further requests. The goal is to see whether the door exists, not to test the lock.

What can I do if I can’t turn TR-069 off?

On ISP-supplied equipment, TR-069 is usually set by the ISP and not shown in the customer settings. That leaves a few practical options, from least to most effort:

  1. Ask the ISP whether 7547 on your line is filtered so that only its ACS can reach it. After 2016 many ISPs added exactly this filtering, so the answer may simply be yes.
  2. Keep firmware updates on. TR-069 is also how many ISPs deliver firmware, so disabling it, where that is possible, can leave known flaws unpatched. The trade-off is real; the filter in step 1 is usually the better fix.
  3. Put your own router behind the ISP device. If the ISP box can run in bridge or modem mode, your own router holds the public address and the ISP device’s management port is no longer what the internet sees. If it cannot, your own router still separates your LAN from the ISP-managed device.
  4. Use your own modem or router, where the ISP allows it, if you want full control and accept responsibility for its updates.

Why a dynamic, often-shared IP changes what a result means

Many home routers running TR-069 also sit behind an ISP-assigned address that changes periodically or is shared across multiple customers via Carrier-Grade NAT — see our CGNAT guide for what that means in practice. That doesn’t make an exposed management port safe, but it does mean a single point-in-time scan of “the internet” for port 7547 may be hitting a different physical router each time it checks the same IP, which is part of why ISP-side firewalling, not just hoping nobody finds your address, is the durable fix. If you’re tracking exposure across more than one site or connection, using port check to inventory IoT and OT device exposure covers turning this same check into a recurring record. This guide is part of AdminKit’s machine protocols cluster of device and protocol exposure guides.

This guide is general information, not a security audit of your specific network — see our disclaimer for what AdminKit's guidance does and does not commit to.

Frequently asked questions

What does TR-069/CWMP actually do?

CWMP (CPE WAN Management Protocol, commonly called TR-069 after the technical report that defines it) lets an ISP's Auto-Configuration Server remotely push firmware updates, change settings, and diagnose problems on the router or modem they provided you, without a technician visiting or you calling support for every change.

Is having TR-069 enabled on my router automatically a problem?

Not by itself — it's a legitimate, widely used management protocol, and a well-implemented ACS connection restricted to your ISP's own management network is a normal, reasonable thing to have running. The 2016 incident happened because specific router models mishandled a related command (TR-064) insecurely, and the management port was reachable from the broader internet, not just the ISP's network.

What actually happened in the 2016 Deutsche Telekom incident?

A Mirai botnet variant scanned the internet for devices listening on port 7547 and sent a crafted TR-064 command that could inject and execute code on vulnerable routers. Deutsche Telekom's specific routers weren't vulnerable to the code-injection itself, but the sheer volume of malicious scan traffic crashed roughly 900,000 of them anyway, as reported at the time by security outlets covering the incident.

How do I know if port 7547 is reachable from outside my ISP's network?

AdminKit's [port check](/tools/port-check/) tool can test whether a specific port on your public IP responds from outside, which is the direct way to find out whether your TR-069 management interface is reachable more broadly than it should be.

Can I just turn TR-069 off myself?

Usually not — on most ISP-provided routers, TR-069 management is configured by the ISP and isn't exposed as a setting you can toggle from the router's own admin page, since the ISP relies on it to manage the device remotely. If you're concerned about exposure, ask your ISP directly whether access is restricted to their management network, and consider replacing an ISP-managed router with your own hardware if you want full control.

Did the 2016 incident get fixed?

Yes — in the immediate aftermath, affected ISPs including Deutsche Telekom restricted public access to port 7547 and pushed firmware patches, and the underlying router vulnerability was catalogued and addressed by the affected vendors. The incident is widely cited as the reason most ISPs now firewall port 7547 from the general internet rather than leaving it openly reachable.

Want AdminKit to watch this for you?

Free checks tell you the state today. Protective DNS and monitoring tell you when it changes.

Huy Nguyen (David), Founder and CEO of Netbase JSC

Talk to an expert

Huy Nguyen (David), Founder & CEO, Netbase JSC

Ask whether this fix applies to your setup, or get a second pair of eyes on your DNS and mail records.