Closed-Loop Containment Assurance in Public Clouds: Compromise-Domain-Aware Enforcer Selection and Scoped Isolation Verification

Abstract Containing a compromised workload is a management action in zero trust clouds, usually recorded as complete once the interface acknowledges the request. The acknowledgement does not show that traffic stopped, and the workload may revert the policy that isolates it. Live measurements confirmed that, as documented, tracked flows survive an isolation security group; a configuration analysis reported no path, and flow records at one interface did not separate two filtering layers. We propose a containment loop built on one compromise model computed from placement, management planes and access-policy reachability; the evaluated controller follows role assumption only. The model decides which evidence may trigger containment, which enforcers the target cannot revert, and which observations may confirm the result. An incident ends EFFECTIVE for a stated scope within the virtual network, or UNRESOLVED; disruptive actions await approval. We evaluated the loop in an emulator of the measured semantics on 10,000 triggered incidents with assumed frequencies. With every established flow tracked and no credentials revoked, containment that trusted the acknowledgement was reported falsely in 75.6% of incidents with a security group alone and in 34.0% with access control lists added, both within one standard error of closed-form expectations. Verification without enforcer selection gave 5.75%. The loop gave 0.12%, only when an injected access-analysis miss coincided with late reversion, which the emulator admits by construction. The price was a median of 11 minutes per declaration and an instance stop in 30% of incidents; in one integrated set, the trigger missed 15% of real bypasses.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-17
DOI
https://doi.org/10.21203/rs.3.rs-11754741/v1
Primary Topic
Security and Verification in Computing
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Closed-Loop Containment Assurance in Public Clouds: Compromise-Domain-Aware Enforcer Selection and Scoped Isolation Verification

Kwangmo Kim
3 links
Research Square
Security and Verification in Computing
preprint

Closed-Loop Containment Assurance in Public Clouds: Compromise-Domain-Aware Enforcer Selection and Scoped Isolation Verification

Kwangmo Kim
preprint en

Abstract

Abstract Containing a compromised workload is a management action in zero trust clouds, usually recorded as complete once the interface acknowledges the request. The acknowledgement does not show that traffic stopped, and the workload may revert the policy that isolates it. Live measurements confirmed that, as documented, tracked flows survive an isolation security group; a configuration analysis reported no path, and flow records at one interface did not separate two filtering layers. We propose a containment loop built on one compromise model computed from placement, management planes and access-policy reachability; the evaluated controller follows role assumption only. The model decides which evidence may trigger containment, which enforcers the target cannot revert, and which observations may confirm the result. An incident ends EFFECTIVE for a stated scope within the virtual network, or UNRESOLVED; disruptive actions await approval. We evaluated the loop in an emulator of the measured semantics on 10,000 triggered incidents with assumed frequencies. With every established flow tracked and no credentials revoked, containment that trusted the acknowledgement was reported falsely in 75.6% of incidents with a security group alone and in 34.0% with access control lists added, both within one standard error of closed-form expectations. Verification without enforcer selection gave 5.75%. The loop gave 0.12%, only when an injected access-analysis miss coincided with late reversion, which the emulator admits by construction. The price was a median of 11 minutes per declaration and an instance stop in 30% of incidents; in one integrated set, the trigger missed 15% of real bypasses.

Research Square
Soongsil University (KR)
Peace, justice, and strong institutions
Security and Verification in Computing
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon