Data Center Network Multiplexed Packet Spraying Switch Fabric
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data center networks face limitations in efficiently utilizing switch fabric bandwidth and providing full mesh interconnectivity between servers, often resulting in suboptimal performance due to reliance on single paths for packet transmission, which can lead to bottlenecks and reduced reliability.
Innovation Solution
The implementation of a data center network system that utilizes multiple parallel data paths within the switch fabric for packet transmission, employing access nodes, permutation devices, and core switches to enable full mesh interconnectivity, allowing packets to be sprayed across multiple paths and reordered for delivery, thereby increasing bandwidth utilization and reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If single path forwarding is used through the switch fabric, then packet transmission is simple and deterministic, but switch fabric utilization is low and bottlenecks occur
Solution Approach 1:
The patent segments the single packet flow into multiple parallel paths through the switch fabric. Instead of forwarding all packets of a flow through one deterministic path, the system divides the flow across multiple paths using packet spraying techniques, thereby increasing switch fabric utilization while maintaining manageable complexity through structured segmentation strategies.
Solution Approach 2:
The patent introduces an additional dimension to packet forwarding by utilizing multiple path dimensions simultaneously. Rather than relying on a single forwarding dimension (one path), the system employs multi-dimensional path selection and packet distribution across parallel routes, increasing throughput while managing complexity through structured multi-dimensional routing protocols.
2Reliability
If multiple parallel data paths are used for packet transmission, then switch fabric utilization increases and reliability improves, but packet reordering complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-establishing path selection criteria and packet tagging mechanisms before packets traverse the multiple paths. Packets are pre-marked with path identification information and reordering requirements, enabling the receiving end to efficiently reorder packets without complex real-time analysis, thus improving reliability while controlling reordering complexity.
Solution Approach 2:
The patent implements feedback mechanisms where the receiving end monitors packet arrival status and path conditions, then adjusts packet reordering strategies accordingly. This feedback-driven approach enables dynamic adaptation to varying network conditions, ensuring reliable delivery while optimizing reordering complexity based on actual traffic patterns and path performance.
3Productivity
If full mesh interconnectivity is implemented between servers, then network efficiency increases, but network complexity and cost increase
Solution Approach 1:
The patent applies universality by designing switch fabric components and access nodes that can handle multiple functions: single-path forwarding, multi-path packet spraying, dynamic path selection, and packet reordering. These universal components can adapt to different interconnectivity requirements without requiring separate dedicated infrastructure for each function, thereby achieving full mesh efficiency while controlling architecture complexity.
Solution Approach 2:
The patent implements dynamic path selection and packet routing strategies that adapt to changing network conditions. Rather than static full mesh connections, the system dynamically adjusts packet distribution across available paths based on load, failure conditions, and performance metrics, achieving high network efficiency while reducing complexity through adaptive rather than fixed architecture.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach significantly enhances switch fabric utilization, providing non-blocking and drop-free full mesh interconnectivity, enabling higher network efficiency and reliability even in massive data centers with thousands of servers, while reducing latency and energy consumption.
Implementation Method 1
Each of the permutation devices is configured such that communications received from input ports are permuted across the output ports based on wavelength so as to provide full-mesh connectivity between the edge-facing ports and the core-facing ports without optical interference
Data Source
AI summary
A network system for a data center is described in which a switch fabric provides interconnectivity such that any servers may communicate packet data to any other of the servers using any of a number of parallel data paths. Moreover, according to the techniques described herein, edge-positioned access nodes, permutation devices and core switches of the switch fabric may be configured and arranged in a way such that the parallel data paths provide single L2/L3 hop, full mesh interconnections between any pairwise combination of the access nodes, even in massive data centers having tens of thousands of servers. The access nodes may be arranged within access node groups, and permutation devices may be used within the access node groups to spray packets across the access node groups prior to injection within the switch fabric, thereby increasing the fanout and scalability of the network system.