NIST 800-53 REV 5 • SYSTEM AND COMMUNICATIONS PROTECTION
SC-3 — Security Function Isolation
Isolate security functions from nonsecurity functions.
Supplemental Guidance
Security functions are isolated from nonsecurity functions by means of an isolation boundary implemented within a system via partitions and domains. The isolation boundary controls access to and protects the integrity of the hardware, software, and firmware that perform system security functions. Systems implement code separation in many ways, such as through the provision of security kernels via processor rings or processor modes. For non-kernel code, security function isolation is often achieved through file system protections that protect the code on disk and address space protections that protect executing code. Systems can restrict access to security functions using access control mechanisms and by implementing least privilege capabilities. While the ideal is for all code within the defined security function isolation boundary to only contain security-relevant code, it is sometimes necessary to include nonsecurity functions as an exception. The isolation of security functions from nonsecurity functions can be achieved by applying the systems security engineering design principles in [SA-8](#sa-8) , including [SA-8(1)](#sa-8.1), [SA-8(3)](#sa-8.3), [SA-8(4)](#sa-8.4), [SA-8(10)](#sa-8.10), [SA-8(12)](#sa-8.12), [SA-8(13)](#sa-8.13), [SA-8(14)](#sa-8.14) , and [SA-8(18)](#sa-8.18).
Practitioner Notes
Security functions — things like access control checks, encryption, and audit logging — need to run in their own protected space, isolated from regular application code. If an attacker compromises a normal application, they should not be able to tamper with your security mechanisms.
Example 1: Use Windows Credential Guard, which runs the LSASS process in a virtualization-based security container. Even if malware compromises the OS kernel, it cannot steal cached credentials because they are isolated in a separate security domain.
Example 2: Deploy a dedicated SIEM server (like Splunk or the Elastic Stack) on a hardened, isolated network segment. Audit logs flow one-way into the SIEM, and even if an attacker compromises production servers, they cannot reach or tamper with the security monitoring infrastructure.