1. Why exposed remote access services are a risk
Some protocols designed for administration or file sharing are often reachable from the internet without anyone remembering having opened them. RDP, the protocol that allows a computer to be controlled remotely, is the most frequent case. SMB is used to share files across a network, while Telnet provides text-based access to systems without encrypting the communication. On the internet these services receive automated access attempts continuously. A weak password or an unpatched system is all it takes for one automated attempt to succeed.
The risk depends not only on the protocol but on how it is exposed: ports open to the outside, weak passwords, outdated versions and no second verification. The work is to understand which services are exposed, why and who uses them, then replace them with more controlled access without stopping those who work. Risk falls fastest when you start with the most exposed services.
2. Prerequisites: inventory and verification from the outside
You cannot close what you cannot see. The first step is an inventory of the services reachable from outside, produced with a check that simulates an external observer, ideally with your supplier's help. The result must list addresses, ports and services, together with the system each one belongs to. If the inventory is incomplete, repeat it after a few weeks: configurations change and a forgotten service can reappear.
Alongside the inventory, gather information about actual use: who connects to these services, from where and how often. Firewall and access system logs are the main source. A service used every day is not closed in the same way as a forgotten one, and that difference must be recorded. Include in the inventory the date of last use, so that genuinely unused services emerge without debate.
- List of services reachable from the internet, with address, port and system.
- An owner for each system and for each rule that exposes it.
- Actual use of the service, taken from logs and confirmed by users.
- List of people who connect remotely, and their devices.
3. How to close without stopping work
Closure should be done as a handover, not as a sudden block. Each exposed service has a replacement that must be ready before the exposure is removed. The most common replacement is access routed through a VPN or an identity-based access method, with multi-factor authentication. The replacement must be designed with the user who will use it, not only with the IT manager.
- For each exposed service, identify the user who uses it and the activity they carry out.
- Prepare a replacement access route and verify it on a test case.
- Communicate the date of the change to the people involved, with instructions for the new access.
- Restrict the old access to the people who still use it, and only from known sources.
- Close the external port and verify from outside that the service no longer responds.
- Keep the documentation of the change and the verification, with date and owner.
4. Authentication and updates for services that remain
Some services cannot be closed immediately, because an application requires them. In this case, reduce them to the minimum: access only through a controlled channel, only named users, no shared passwords and multi-factor authentication wherever the service allows it. Where this is not possible, document the reason and the date by which the service will be replaced. Also verify that the system is up to date, because known vulnerabilities in these services are actively searched for.
For Telnet the practical guidance is clearer: instead of plaintext access, use a protocol that encrypts the communication, such as SSH, and disable Telnet wherever possible. For SMB, disable the oldest versions of the protocol and never expose shared folders to the internet. These recommendations should be confirmed with your supplier based on the systems you use.
5. Common mistakes
The most frequent mistake is closing a service without knowing who uses it: the next day a supplier can no longer intervene on a device, or a departmental application stops synchronising. The second is treating closure as a one-off task. New services are published, often for a test or an urgent need, and remain open. A simple rule is needed: no external port without a purpose, an owner and a review date. A closed service must also be removed from the documentation, otherwise someone will reactivate it thinking it is still planned.
Another mistake is moving the exposure to a different port. Changing the port does not close the service, because scanning tools find it anyway. Finally, be careful with exceptions requested by a supplier: if the request is urgent, the exception must be time-limited and recorded, otherwise it stays forever. Write the expiry date in the register and check it when it arrives, without renewing it automatically.
6. How to confirm the work has succeeded
After each closure, repeat the external check on the same service. The expected result is that the port no longer responds. Then check that the replacement works for the people who need it, through a real test rather than a declaration. The work is considered closed only when both conditions are true. If either condition fails, the work remains open and must be resumed from the same point.
Keep a register with the date of verification, the owner and the result. Repeat the external check at regular intervals, because a new rule or a new device can re-expose a service you had closed. Periodic verification matters more than a one-off check. Schedule it as a fixed appointment in the calendar of whoever manages the network, with an owner who receives the result.
7. How to start
This week, ask your supplier or IT manager to produce the list of services reachable from the internet. If that cannot be obtained within a few days, at least ask for the firewall rules that forward ports to the inside. Do not close anything yet: first you need to know what exists and who uses it. If you find services that nobody can explain the use of, mark them as to be checked, without deciding yet.
Then identify the riskiest service, usually an administrative access left open to the internet, and plan its replacement with an agreed date. When the replacement is ready and tested, close the exposure and document the result. This is the safest way to start without interrupting work. Communicate the result to those who use the service, with the new access method already working.
Frequently Asked Questions
Is RDP protected by a strong password secure?
Not secure enough to be exposed to the internet. A strong password reduces the risk of guessing, but it does not protect against vulnerabilities in the service, repeated automated attempts or credentials stolen elsewhere. The service should be reachable only through a controlled channel, with multi-factor authentication.
Can I close everything straight away and see who complains?
It is a strategy that causes avoidable damage. Closing without an inventory brings dependencies to light at the worst moment, when an urgent intervention is needed. Proceed with the list of users and a tested replacement, communicate the date and then close. That way the work stays protected and the closure does not become an emergency.
Can the supplier of our management software ask for external access?
It can, and sometimes it is unavoidable. Before granting access, ask which port it uses, from which addresses, with what authentication and for which activities. Prefer a mediated access route, such as a VPN or identity-based access, and set an expiry date for every exception. Also note who at the supplier holds the credentials.
Does changing the port help at all?
Almost nothing from a security point of view. Scanning tools also check non-standard ports, and a moved service remains exposed. Moving it may reduce the noise of automated attempts, but it does not replace closure or access control.
Who should approve the closure of a service?
The system owner and the internal IT manager, with communication to the people who use it. If the service is managed by a supplier, the supplier will carry out the change, but the decision must be recorded. An approved and documented closure avoids unnecessary discussion when someone complains about an outage.