As a technology that was built into Windows to enable the remote of computers, RDP can be an easy fix to such problems, but can also become a major weakness for organisations if deployed insecurely.
RDP a serious problem made worse
The RDP protocol is a frequent target for credential stuffing and other brute-force password guessing attacks that rely on lists of common usernames and password combinations or on credentials stolen from other sources. Some cybercriminals even specialise in selling hacked RDP credentials as a commodity on the underground market to other hackers who use them to deploy ransomware and cryptominers or to engage in more sophisticated attacks that can lead to the theft of sensitive data and more extensive network compromises.
“McAfee has noticed an increase in both the number of attacks against RDP ports and in the volume of RDP credentials sold on underground markets,” researchers from security firm McAfee said in a new report.
The company notes that the number of RDP ports exposed to the internet has grown from around 3 million in January to more than 4.5 million in March. More than a third of them are in the US and another third are in China. More than half of the machines with exposed RDP ports are running some version of Windows Server, but around a fifth run Windows 7, which is no longer supported and does not receive security updates. That is a concern because in addition to often being configured with weak passwords, RDP has also seen its share of vulnerabilities and exploits over the years.
Around half of all RDP credentials sold on the underground market are for machines in China, followed by Brazil, Hong Kong, India and the US. The number of credentials for US-based RDP hosts is fairly low, at 4% of the total, but McAfee believes this is likely because the hackers who sell them do not publish their entire lists and hold the more valuable credentials and hosts for themselves or more private and select sales.
A surge in RDP attacks
According to another recent report from VPN service provider Atlas VPN, starting on 10th March, the number of RDP attacks have spiked significantly in the US, Spain, Italy, Germany, France, Russia and China. This seems to correlate with the beginning of the population movement restrictions and lockdowns enforced around the world in response to the Covid-19 pandemic.
“The attacks peaked on 7th April 2020, with a total number of 1,417,827 attacks,” the company said in its report. “Comparing the period of 9 February through 9th March 2020, to 10th March through 10th April 2020, the RDP attacks in the US jumped by 330%.”
Between 10th March and 15th April, the company recorded 148 million RDP attacks around the world. More than 32 million of them were detected in the US, or almost 900,000 attacks per day on average.
“These attacks systematically attempt numerous username and password combinations until the correct one is found,” the company said. “A successful attack gives the cybercriminal remote access to the target computer or server in the corporate network.”
First, exposing RDP directly to the internet is bad security practice, even with good credential hygiene, digital certificates and two-factor authentication. Slow patching can always lead to servers being compromised through an RDP vulnerability. RDP should always be accessible only through a secure VPN connection to the corporate network or through a zero-trust remote access gateway like LoughTec’s Connect Gateway.
LoughTec recommends the following best practices:
- Do not allow RDP connections over the open internet.
- Use complex passwords as well as multi-factor authentication.
- Lock out users and block or timeout IPs that have too many failed logon attempts.
- Use an RDP gateway.
- Limit domain admin account access.
- Minimise the number of local admins.
- Use a firewall to restrict access.
- Enable restricted admin mode.
- Enable Network Level Authentication (NLA).
- Ensure that local administrator accounts are unique and restrict the users who can logon using RDP.
- Consider placement within the network.
- Consider using an account-naming convention that does not reveal organisational information.