Using just-in-time access to reduce cloud security risk

Excessive privileges are a continuing headache for security professionals. As more organizations migrate assets to the cloud, users with excessive permissions can expand the blast radius of an attack, leaving organizations open to all sorts of malicious activity.

Cloud environments rely on identity as the security perimeter, and identities are mushrooming and making “identity sprawl” a serious challenge. Users often have multiple identities that span many resources and devices, while machine identities —used by apps, connected devices and other services—are growing at an accelerated pace.

This becomes a problem if an attacker manages to compromise an identity, allowing them to gain a foothold in the environment and exploit those privileges to move laterally throughout the cloud environment — or even escalate permissions to do even more damage across many other assets and resources.

One way to address the large attack surface and unnecessary risk in the cloud is to implement just-in-time (JIT) privileged access. This approach limits the amount of time an identity is granted privileged access before they are revoked. Even if an attacker compromises credentials, it may only have privileged access temporarily or not at all. This is a critical defense mechanism.

Simply put, JIT grants privileged access only temporarily and revokes it once the related task is completed. JIT builds on a least-privilege framework to include a time factor, so users only have access to those resources they need to carry out their functions, and only while they are performing those functions. That said, excessive privileges should, by default, be eliminated wherever possible.

“Right-sizing permissions” has become a buzzword for security professionals, but it’s a challenge. Enforcing the kind of granular permissions management necessary for good cloud security manually—going back and forth trying to determine which privileges are called for and what are the minimal escalations that can get the job done — can be time-consuming and frustrating for both users and security teams.

Organizations have reason to worry. As the annual Verizon Data Breach Investigations Report notes time and again: credentials can be the weak link in any network. The most recent report noted the use of stolen credentials has grown about 30% in the last five years. Since a large share of breaches can be traced back to credential theft and abuse, limiting the potential scope of account compromise will have an outsized effect on improving security.

How to implement JIT access

Deploying JIT access begins with gaining a clear view of who users are, what privileges they have and what privileges they need, including whether they are human and machine identities. Is the user an engineer or developer, an administrator or security staff?
Work can’t stop while a user waits to be validated. This is where automation can provide a workable system to provision temporary privileges and revoke them once they’re not necessary.

A few best practices can help security teams implement automated JIT:

• A self-service portal: Security staff get a bad rap as creators of user friction, so any tool that can smooth out workflows is a good thing. A self-service portal can reduce friction by allowing users to request elevated privileges and tracking the approval process. This cuts back on delays and requests that fall through the cracks, while also enabling automated permissions management, which in turn reduces cloud attack surface and leads an audit trail for monitoring activity.
• Automate policies for low-risk requests: Simple requests involving low-risk activity, such as work in non-production environments, can be automated with policies that approve requests for a limited time and without human intervention.
• Define owners for each step of the process: Automation should not equal relinquishing control of business processes. It needs to be monitored to ensure unintended actions do not occur. Each step of the process —reviewing requests, monitoring implementation, and revoking privileges—must be assigned an owner and more complex and sensitive requests should be reviewed and approved by a human, when necessary.

By implementing JIT, security teams can move closer to achieving a least-privilege model and implementing zero trust security. Automation can make this possible by speeding up the process of granting and revoking permissions as necessary, without creating more work for security teams that are already stretched thin, or friction for users that impacts their agility and efficiency.