Nearly half of Windows servers could expose entire network if privileged accounts stolen, study warns
- 13 November, 2015 08:37
Fully 40 percent of Windows servers have privileged accounts that could be used by attackers to gain unimpeded access to the majority of other network hosts, CyberArk has reported after an audit of thousands of Windows hosts running on a range of different corporate networks.
Some using the CyberArk Discovery & Audit (DNA) tool, which provided a detailed assessment of environments ranging from small businesses with a few dozen Windows hosts to large enterprises with thousands of Windows hosts.
Among CyberArk's evaluations were a measurement of how many other organisational servers are accessible from a given privileged account; a 'high-risk host' was defined as being one with credentials that enable access to 80 percent or more of the network's other Windows hosts.
“An attacker that compromises the source host can readily steal additional credentials from it and use them to gain access to the target hosts,” the report warns.
Exposure to unmanaged and improperly configured privileged accounts remains a significant factor for companies of all sizes, with those consuming cloud services facing even greater risks. The importance of careful privileged-account management has also grown as companies like Telstra develop large internal repositories of data that offer great appeal to cybercriminals, and others, like Vodafone, reconsider the access they are providing to third parties.
The new CyberArk analysis confirmed that once privileged credentials are exploited by outsiders, other networked hosts are likely to follow quite quickly. The largest cohort of companies – 11 of the 51 (22 percent) – had between 51 percent and 60 percent high-risk networked hosts. The mean for all companies was 40 percent, while the median value was 36 percent.
These figures represent significant vulnerabilities from privileged accounts: “What makes this simple metric so interesting is that it reflects the process that attackers follow to reach their ultimate targets,” the analysis notes.
“An attacker often starts an attack through phishing or other means to gain access to a user account on a single host... then can use those privileges to jump from host to host. Networks with a high value are making it much easier for attackers to reach their goals.”
The analysis found no correlation between network size and the percentage of high-risk hosts, with at-risk host measurements evenly distributed on a range of differently-sized networks.
“This means that every Windows network, no matter how large or small, could potentially be compromised by attackers through theft of privileged credentials,” the analysis warns.
Overall, 19 networks were classified as high exposure and 23 classified as medium exposure; only 7 were classified as having low exposure to privileged-account compromise.
CyberArk's analysis also examined the proliferation of high-risk accounts – defined as an account that can enable direct access to more than 80 percent of the network's other hosts – and grouped them according to whether they are used to access systems by people (as in 23 of the organisations), automated processes (in 11 organisations), or both (in 17 organisations).
Different usage profiles inform the remediation steps necessary for organisations to minimise their exposure to those profiles, the analysis warned: accounts used by people, for example, should be limited to privileged local accounts rather than privileged domain accounts and administrators should consider minimising their exposure to credential compromise by using one-time passwords that block each login.
Other mitigation strategies include zoning – limiting account access to a small subset of network hosts – and implementing limited privileged domain accounts that have high privileges on a specific host but limited privileges on other hosts.
For accounts that are used in automated accounts, organisations can minimise their risk by ensuring their systems don't hardcode access credentials and instead manage them in a centralised credentials repository.
“This greatly lowers the threat of an attacker retrieving the embedded credentials, and it also enables the organisation to regularly rotate the passwords for these accounts,” the analysis noted. Hard-coded passwords, by contrast, are often “used for years because of the overhead involved in modifying the software just to change a password.”
[N.B.: It might be worth embedding somewhere in this story]