Network Managing Device for Fat Tree Switch Error Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In large-scale InfiniBand network configurations, manual errors in connecting first and second stage switches can lead to erroneous connections, reducing network efficiency and signal transmission performance due to path competition, requiring extensive manual correction and increasing construction time.
Innovation Solution
A network managing device with an initial address allocating unit, connecting state acquiring unit, erroneous connection detecting unit, and address changing unit that automatically detects and corrects erroneous connections by reallocating node addresses to prevent path competition in a Fat Tree configuration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If manual connection operation is used to connect first stage switch and second stage switch, then connection flexibility is maintained, but connection accuracy deteriorates due to high possibility of erroneous connection
Solution Approach 1:
The system performs self-diagnosis by automatically detecting erroneous connections through address allocation conflicts. The network managing device autonomously identifies connection errors without requiring manual inspection, and automatically reallocates addresses to resolve the errors, making the system self-correcting.
Solution Approach 2:
The system implements feedback mechanisms by monitoring address allocation results and detecting conflicts that indicate erroneous connections. Based on this feedback, the system automatically reallocates addresses to correct the connection errors, creating a closed-loop control system that continuously optimizes connection accuracy.
2Manufacturing precision
If erroneous connection is detected and corrected manually, then connection accuracy is improved, but construction time is prolonged due to extensive manual correction work
Solution Approach 1:
The system automatically detects and corrects connection errors through autonomous address allocation and conflict detection. When erroneous connections are identified, the system self-corrects by reallocating addresses without requiring manual intervention, thereby eliminating the time-consuming manual correction process while maintaining high connection accuracy.
Solution Approach 2:
The system performs preliminary address allocation before finalizing connections, and proactively detects potential conflicts that indicate erroneous connections. By addressing connection errors during the address allocation phase rather than after connection completion, the system prevents time losses associated with post-connection debugging and correction.
3Ease of operation
If standard routing is carried out in erroneous connection state, then routing simplicity is maintained, but network performance deteriorates due to path competition
Solution Approach 1:
The system monitors address allocation results to detect conflicts that indicate erroneous connections affecting path competition. Based on this feedback, the system automatically reallocates addresses to eliminate conflicts, thereby resolving path competition issues while maintaining routing simplicity through automatic correction rather than complex manual routing adjustments.
Solution Approach 2:
The system changes address allocation parameters dynamically to resolve conflicts caused by erroneous connections. By reallocating addresses differently than the initial allocation, the system eliminates path competition without requiring changes to the underlying routing logic or network topology, thus maintaining routing simplicity while improving performance.
Data Source
AI summary
An initial address allocating unit 11 allocates a predetermined address to each of the nodes 21 connected to respective first stage switches 22. A connecting state acquiring unit 12 acquires a connecting state from each of the first stage switches 22, each of second stage switches 23 and each of third stage switches 24. An erroneous connection detecting unit 13 compares the connecting state acquired by the connecting state acquiring unit 12 with a prestored predetermined connecting state, thereby detecting an erroneous connection. An address changing unit 14 changes an address of the node 21 allocated by the initial address allocating unit 11 to satisfy a first predetermined condition based on the connecting state, when the erroneous connection is detected by the erroneous connection detecting unit 13.