An approach for addressing (e.g., preventing) detected network intrusions in a virtualized/networked (e.g., cloud) computing environment is provided. In a typical embodiment, users may group components/systems of an environment/domain according to a range of security sensitivity levels/classifications. The users may further configure rules for responding to security threats for each security sensitivity level/classification. For example, if a “highly dangerous” security threat is detected in or near a network segment that contains highly sensitive systems, the user may configure rules that will automatically isolate those systems that fall under the high security classification. Such an approach allows for more granular optimization and/or management of system security/intrusion prevention that may be managed at a system level rather than at a domain level.