Skip to main content

Advances, Systems and Applications

Table 4 Comparison (RS: Responsible stakeholder, CP: Cloud Provider, CC: Cloud Consumer, P: Proactive, R: Reactive)

From: Handling compromised components in an IaaS cloud installation

Approach

RS

P/R

Service impact

Implementation/Enforcement difficulties

Dependencies

Filtering in the messaging server

CP

R

Platform components may never receive an expected message.

Unless deployed in distributed mode, can become a bottleneck.

Messaging server

Filtering in each component

CP

R

Platform components may never receive an expected message.

All components should be modified to support it.

Platform components

Disabling services

CP

R

Healthy components can become inaccessible. Losing control over instances managed by disabled components.

-

Platform interfaces

Replicating services

CP

P

Services should be replicated based on requirement and performance analysis of the environment.

-

Platform components

Disinfecting infected components

CP

R

Healthy components can become inaccessible. Losing control over instances managed by disabled components.

Configuration management tools and cloud platform interfaces should be deployed and configured.

Configuration management tools, Platform interfaces

Removing instances from the project VLAN

CP CC

R

The instance won’t be accessible for the consumer and its services.

Highly dependent on the OpenStack VLANManager networking mode.

Platform components

Disabling live migration

CP

R

Consumer experiences lower QoS.

-

Platform interfaces

Quarantining instances

CP CC

R

Quarantined instances won’t be accessible for the consumer.

Implementing this solution requires a lot of effort as discussed briefly in[10].

-

Disinfecting an instance

CP CC

R

-

A framework for analyzing VM images and disinfecting running instances must be developed

Platform interfaces

Migrating instances

CP CC

R

Consumer may experience lower QoS.

The cloud environment should consist of distributed and independent zones.

Platform interfaces

Component authentication

CP

P

Small overhead for all communications.

Developing a system for managing components certificates and identity.

Messaging server and identity services

No new worker policy

CP

P

-

Developing a policy manager component

Messaging server and policy manager

Trust levels and timeouts

CP CC

P

Lower QoS for non-critical use-cases Lower resource volume for critical use-cases

High complexity

Platform interfaces, and scheduler component