Virtual Machine Introspection via Hypervisor Intermediary

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current virtual machine monitoring techniques face challenges in efficiently and minimally impacting system performance, especially when monitoring multiple virtual machines generating diverse software behaviors simultaneously, and require robust and flexible solutions for introspection without the need for separate agents.

Innovation Solution

The implementation of a virtual machine introspection framework using a modified Xen hypervisor with a robust API, allowing for selective interception of system calls and events, and leveraging operating-system-specific libraries to analyze software behavior without significant overhead, enabling monitoring of multiple VMs with minimal performance impact.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If virtual machine introspection is implemented to monitor VM activities, then monitoring capability is improved, but system performance impact increases

Engineering Contradiction:
Improvemonitoring capabilityVSAvoidsystem performance
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The hypervisor acts as an intermediary between the host and guest VMs, implementing introspection facilities that monitor VM activities without requiring agents inside the VMs. The hypervisor intercepts system calls and traps at appropriate points to gather monitoring data while maintaining performance through efficient hardware-assisted virtualization mechanisms.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The system dynamically adjusts introspection intensity and monitoring parameters based on workload characteristics. By changing monitoring parameters such as trap frequency, data collection granularity, and event filtering thresholds, the system optimizes the balance between monitoring capability and performance impact under different operational conditions.

Inventive Principle:
Principle #35Parameter changes

2Reliability

If comprehensive monitoring of multiple VMs is implemented, then monitoring coverage is improved, but system complexity increases

Engineering Contradiction:
Improvemonitoring coverageVSAvoidsystem complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The hypervisor implements a universal introspection framework that can monitor multiple different VM types and operating systems through a single unified mechanism. The same hypervisor code handles diverse monitoring needs for various VMs, reducing overall system complexity while maintaining comprehensive coverage.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The monitoring system is segmented into modular components including event capture, data collection, analysis, and presentation layers. This segmentation allows independent optimization and management of each function, making the overall complex system more manageable and maintainable while expanding monitoring capabilities.

Inventive Principle:
Principle #1Segmentation

3Reliability

If agent-based monitoring is used, then monitoring functionality is improved, but installation complexity and overhead increase

Engineering Contradiction:
Improvemonitoring functionalityVSAvoidinstallation complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The hypervisor serves as the intermediary that provides monitoring functionality without requiring agents to be installed inside guest VMs. All monitoring is performed from the host side through the hypervisor, eliminating the complexity of agent deployment, updates, and maintenance while maintaining full monitoring functionality.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The VMs themselves are monitored through their own execution traces and system calls without external assistance. The hypervisor automatically captures and analyzes VM activities as they naturally occur, eliminating the need for additional software components inside the VMs and simplifying the overall system architecture.

Inventive Principle:
Principle #25Self-service

Data Source

PatentUS9619346B2Virtual machine introspection facilities
Publication Date: 2017.04.11 ASSURED INFORMATION SECURITY
  • US9619346B2 patent drawing
  • US9619346B2 patent drawing
  • US9619346B2 patent drawing

AI summary

Run-time, event-driven virtual machine introspection of the target guest virtual machine is facilitated as described herein. A component can specify events that are of interest to the component for introspection of a target guest virtual machine of a hypervisor. The hypervisor detects an introspection event generated by a target guest virtual machine and determines whether the introspection event is of interest for handling by a component coupled to the hypervisor. If so, the hypervisor alerts the component about the introspection event and provides information associated with the introspection event to the component. The component thereby receives notification of occurrence of the introspection event from the hypervisor and may obtain information associated with the introspection event.