ASSESSING THE IMPACT OF ZERO TRUST ON INCIDENT LOCALIZATION IN CONTAINERIZED MICROSERVICE ENVIRONMENTS

Abstract

In containerized microservice architectures, the compromise of a single service can become a starting point for lateral movement deeper into the infrastructure because internal service interactions often preserve an excessive level of implicit trust. The Zero Trust approach requires service-to-service mutual authentication, least privilege, enforced microsegmentation, and explicit verification of each request; however, publicly available guidance still lacks a formalized method for quantitative comparison of system states before and after such controls are introduced. This paper proposes a model environment with three services, a threat scenario based on a compromised app-service, and three evaluation metrics: impact radius R, defined as the share of reachable access objects; lateral movement depth D, defined as the maximum number of sequential transitions; and policy operational complexity C, defined as the number of security artifacts that must be maintained in an up-to-date state. Model-based calculations show that implementing Zero Trust reduces R from 0.93 to 0.21, decreases D from 2 to 1, and increases C from 2 to 13. For an extended five-service nonlinear topology, the same trend is observed: R decreases from 0.83 to 0.17, D again decreases from 2 to 1, and C rises to 28, indicating that the localization effect remains stable in a more complex architecture. Empirical validation on a local Kubernetes cluster with Calico CNI showed that 17 of 17 access tests were successful in the baseline state, whereas only 6 of 17 remained successful after Zero Trust hardening. These results confirm the model predictions and also demonstrate that complete restriction at the level of individual operations requires L7 authorization, typically provided through a service mesh. Overall, the findings indicate that enforced Zero Trust controls substantially reduce the propagation potential of an attack, improve the architecture’s ability to localize the consequences of compromise, and provide a useful basis for further quantitative security assessment of cloud-native systems. At the same time, this security improvement is accompanied by a noticeable increase in the operational burden associated with maintaining policies, certificates, segmentation rules, and related security artifacts.