Logical Data Flow Mapping for Cluster Isomorphism
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cluster computing systems face challenges in implementing write-once, repetitive use applications that are portable across different configurations due to varying computing resources and lack of abstract programming tools that can account for all permutations, leading to difficulties in ensuring structural compatibility and isolating structural components.
Innovation Solution
The use of directed acyclic graphs (DAGs) with data flow functions, including assertions and transformations, to map logical data flows onto cluster computing systems, employing rules and actions to test and modify the DAGs to conform to target cluster architectures, allowing for iterative conversion into executable physical data flows.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If programming tools are used to implement applications across cluster computing systems, then application functionality can be achieved, but structural compatibility and portability across different cluster configurations cannot be ensured due to varying computing resources and lack of abstraction
Solution Approach 1:
The patent introduces an intermediary layer consisting of a directed acyclic graph (DAG) representation and mapping rules that act as a mediator between the application logic and the underlying cluster computing infrastructure. This intermediary abstracts the heterogeneity of different cluster configurations, allowing applications to be ported across platforms while maintaining structural compatibility. The DAG serves as a universal intermediate representation that can be mapped to various cluster architectures through defined mapping rules.
Solution Approach 2:
The patent employs parameter changes by transforming the application representation from a platform-specific format to a standardized DAG format with specific parameters (nodes, edges, attributes). This transformation allows the same logical data flow to be represented universally and then adapted to different cluster configurations by changing the mapping parameters rather than the core application logic.
2Productivity
If cluster computing systems use multiple stand-alone computing devices with varying resources, then computing power and flexibility are increased, but the complexity of ensuring structural compatibility and reliable execution across devices increases
Solution Approach 1:
The patent segments the complex task of ensuring structural compatibility across multiple devices by breaking it down into manageable components: (1) representing the application as a DAG with specific structural properties, (2) defining mapping rules that operate on individual DAG elements, and (3) systematically applying these rules to verify compatibility. This segmentation transforms an intractable global verification problem into a series of localized, manageable checks.
Solution Approach 2:
The patent creates a universal DAG representation that can serve multiple functions: it represents the application logic independently of the target platform, serves as a basis for automatic mapping to different cluster configurations, and enables systematic verification of structural compatibility. This multi-functional intermediate representation simplifies the complexity of working with heterogeneous computing devices.
3Adaptability or versatility
If programming languages attempt to account for all permutations of cluster resources, then comprehensive coverage is achieved, but the complexity and abstraction level required increases significantly
Solution Approach 1:
The patent introduces dynamics by making the mapping process adaptive rather than static. Instead of requiring programming languages to hard-code all possible resource permutations, the system dynamically generates appropriate mappings based on the actual target cluster configuration. The DAG representation and mapping rules can be applied to any cluster configuration, automatically adapting to the specific resources available without requiring pre-programming of all permutations.
Data Source
AI summary
A system and a method are disclosed for transforming cluster computing resources of a target system to match a user defined logical data flow so data processing between the source and target are functionally equivalent. The source logical dataflow is compatibly mapped to a target directed acyclic graph that represents the cluster computing resources. A series of subgraph assertion and transform operations are applied iteratively until the logical data flow is isomorphic with a directed acyclic graph. The assertion and transform operations are comprised of rules and assertions, which are maintained in a rules registry.