Data Exchange Buffer Virtualization for Network Congestion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional data exchange networks suffer from low efficiency and increased delay due to congestion, buffer overflow, and head-of-line blocking, despite using methods like ECN, PFC, and tail drop.
Innovation Solution
A method where a source node or switching node detects congestion and distributes data blocks across multiple switching nodes with larger buffer capacity, utilizing a virtualized buffer pool to store and manage data blocks, reducing congestion and improving efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional flow control methods (ECN, PFC, tail drop) are used, then congestion is reduced to some extent, but exchange efficiency deteriorates and delay increases
Solution Approach 1:
The patent segments the buffer resource into multiple virtual buffer queues corresponding to different data flows, allowing independent management and scheduling of each flow's data packets. This segmentation enables fine-grained congestion control without affecting overall exchange efficiency, as each flow can be handled separately while utilizing the shared buffer pool effectively.
2Reliability
If buffer capacity of switching nodes is increased, then burst traffic absorption capability is improved, but device complexity and cost increase
Solution Approach 1:
The patent merges the buffer resources of multiple switching nodes into a shared virtual buffer pool, allowing dynamic allocation and sharing of buffer capacity across the network. This approach provides burst traffic absorption capability equivalent to a large buffer without requiring each individual switching node to have large physical buffer capacity, thereby reducing device complexity and cost while maintaining reliability.
3Ease of operation
If data packets are stored in a single buffer queue, then management is simple, but head-of-line blocking occurs and exchange efficiency decreases
Solution Approach 1:
The patent segments the single buffer queue into multiple virtual buffer queues, each corresponding to a specific data flow. This segmentation prevents head-of-line blocking by allowing out-of-order delivery and reassembly of packets from different flows, thereby maintaining ease of buffer management while significantly improving exchange efficiency through parallel processing of multiple flows.
4Reliability
If multiple data blocks are distributed to multiple switching nodes, then buffer capacity is effectively increased and congestion is reduced, but data exchange complexity increases
Solution Approach 1:
The patent implements feedback mechanisms where switching nodes monitor buffer status and congestion conditions, and dynamically adjust data block distribution and forwarding decisions. This feedback-based approach automates the complex data exchange process, allowing the system to adaptively optimize congestion reduction while managing complexity through intelligent control rather than static complex routing rules.
Data Source
AI summary
A data exchange method and apparatus. The method includes: A source node receives flow indication information from a first switching node, where the flow indication information indicates that congestion occurs in a target data flow, and the first switching node is a node in a switching path of the target data flow. The source node sends multiple pieces of write data information and multiple data blocks of the target data flow to multiple switching nodes, where the multiple pieces of write data information indicate the multiple switching nodes to store the multiple data blocks and stop forwarding the multiple data blocks. The multiple switching nodes store the multiple data blocks. The multiple switching nodes receive multiple pieces of scheduling information from a destination node, where the multiple pieces of scheduling information are used to schedule the multiple data blocks.