Compute Node Integrity Checking via Distributed Application Test Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for checking the integrity of compute nodes in networks of communicatively connected compute nodes, such as cloud edge platforms, are inadequate for safety-critical applications due to dynamic configuration and unknown software hardware stacks, leading to potential systematic errors and difficulties in ensuring correct functioning.
Innovation Solution
A method involving the provision of test data to compute nodes, generating test results by applying a distributed application, and checking integrity by comparing these results to expected outcomes, with features like atomic test data sets, temporary state resetting, and signature generation for reliable integrity assessment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If built-in self-tests are used to check hardware integrity, then hardware errors can be detected, but the method is not suitable for dynamic cloud edge platforms with unknown software hardware stacks
Solution Approach 1:
The patent uses test data sets as copies of actual application data to verify compute node integrity. Instead of relying on built-in hardware tests that are tied to specific hardware configurations, the system creates software-based copies of data processing tasks that can be executed on any compute node regardless of its underlying hardware or software stack, enabling universal integrity checking across dynamic cloud edge platforms
Solution Approach 2:
The patent introduces test data sets and reference results as intermediary elements between the compute node and the integrity verification system. These intermediaries enable indirect verification of compute node correctness by comparing test outputs against known reference results, bypassing the need for direct hardware access or knowledge of the compute node's internal software hardware stack
2Adaptability or versatility
If dynamic configuration is used to improve resource efficiency and flexibility, then cloud edge platforms become more adaptable, but integrity checking becomes difficult due to mutual influences and abstraction layers
Solution Approach 1:
The patent segments the integrity verification process into independent components: test data generation, test execution, result collection, and comparison with reference results. This segmentation allows each component to operate independently without being affected by the dynamic configuration of other parts of the system, simplifying integrity checking in complex distributed environments
Solution Approach 2:
The patent changes the verification approach from hardware-level parameters (built-in self-tests) to software-level parameters (test data sets and reference results). This parameter change enables integrity verification to work independently of the underlying hardware configuration, abstraction layers, and dynamic resource allocation, making the system adaptable to changing configurations while maintaining verification capability
3Reliability
If built-in monitoring mechanisms are used in safety-focused computing platforms, then safety-critical errors can be recognized, but these mechanisms are not integrated into cloud edge platform compute nodes
Solution Approach 1:
The patent creates a universal integrity verification method using test data sets that can be applied to any compute node in cloud edge platforms regardless of its specific configuration. The same verification approach works across diverse hardware architectures, operating systems, and software stacks, providing safety-critical error recognition capability without requiring platform-specific integration or modification
Data Source
AI summary
A method for checking the integrity of a compute node which is part of a network of compute nodes communicatively connected to one another, and on which at least one portion of a distributed application is executed. The method includes: providing test data for the compute node; generating test results by applying the at least one portion of the distributed application to the provided test data; and checking the integrity of the compute node by comparing the test results to expected results.