Cloud Manager STONITH Fencing for VM Isolation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In cloud and virtualized environments, existing methods for fencing and STONITH (Shoot The Other Node In The Head) operations are inefficient due to the lack of direct hardware access, leading to potential split brain scenarios and delayed node isolation, especially in telecommunication applications requiring high availability.

Innovation Solution

Implementing a cloud manager with a STONITH/fencing application within the cloud infrastructure, allowing virtual machines to directly communicate and authorize requests for node control within a recovery domain, reducing latency and security risks by eliminating the need for external authorization tokens and minimizing network hops.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional fencing methods using external hardware functions (IPMI) are used in cloud environments, then node isolation can be achieved, but direct hardware access is no longer possible and latency increases

Engineering Contradiction:
Improvenode isolation reliabilityVSAvoidhardware access complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent introduces a cloud manager as an intermediary component that mediates between virtual machines and the fencing mechanism. The cloud manager receives fencing requests from virtual machines, validates them using authorization tokens, and executes the fencing operation by controlling the virtual machine lifecycle. This intermediary approach resolves the contradiction by enabling node isolation without direct hardware access, as the cloud manager translates high-level fencing requests into appropriate virtual machine control operations.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent implements a software-based fencing mechanism that copies the functionality of traditional hardware-based fencing. Instead of directly accessing hardware through IPMI, the system creates a virtualized copy of the fencing capability within the cloud manager. This software abstraction layer replicates the isolation function while adapting it to work within the constraints of virtualized environments where direct hardware access is unavailable.

Inventive Principle:
Principle #26Copying

2Reliability

If cloud managers authorize requests using external tokens, then security is maintained, but network hops increase and latency increases

Engineering Contradiction:
Improveauthorization securityVSAvoidauthorization latency
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent implements preliminary action by pre-configuring authorization tokens within the virtual machine's identity context before fencing operations are needed. The cloud manager establishes trust relationships in advance, where virtual machines are pre-authenticated and their authorization credentials are validated beforehand. This eliminates the need for real-time token verification during critical fencing operations, significantly reducing authorization latency while maintaining security through pre-established trust credentials.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent extracts the authorization token verification process from the external cloud manager infrastructure and embeds it directly within the virtual machine's identity context. By taking out the external authorization dependency and integrating authentication credentials into the virtual machine itself, the system eliminates unnecessary network hops to external authorization services. The cloud manager can validate credentials locally using the embedded identity information, reducing latency while maintaining security.

Inventive Principle:
Principle #2Taking out (Extraction)

3Speed

If virtual machines can directly control other virtual machines, then fencing speed increases, but security risks increase

Engineering Contradiction:
Improvefencing speedVSAvoidsecurity risks
Core Design Contradiction:
SpeedVSObject-affected harmful factors

Solution Approach 1:

The patent implements a feedback mechanism where the cloud manager continuously monitors and validates fencing requests from virtual machines. When a virtual machine initiates a fencing operation, the cloud manager receives the request, validates the initiator's credentials and authorization tokens, verifies the target virtual machine's status, and then executes the fencing operation. This feedback loop ensures that only authenticated and authorized virtual machines can perform fencing operations, maintaining security while enabling direct control for rapid response.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent introduces dynamic authorization where the ability of virtual machines to control each other is not static but dynamically validated based on current system state and pre-established policies. The cloud manager dynamically adjusts authorization decisions based on the specific context of each fencing request, including the identity of the initiator, the target virtual machine, and the current system state. This dynamic approach enables fast fencing operations while maintaining adaptive security that responds to changing conditions.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS10884775B2Methods and apparatus to control a virtual machine
Publication Date: 2021.01.05 NOKIA SOLUTIONS & NETWORKS OY
  • US10884775B2 patent drawing
  • US10884775B2 patent drawing
  • US10884775B2 patent drawing

AI summary

Methods and apparatus, to control a virtual machine, include: sending identification of virtual machines within a recovery domain, receiving a request to control one the virtual machines, authorizing the request, and forwarding the request to a virtual machine manager of the one of the virtual machines. Other methods and apparatus, to control a virtual machine, include: sending a request for identification of virtual machines within a recovery domain directly to a cloud computing device without passing the request through another computing device, determining that one of the virtual machines needs to be controlled, and sending a request to control the one of the virtual machines directly to the cloud computing device without passing the request through another computing device.