Fabric Control Protocol for Data Center Packet Spraying
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data center networks face issues such as fabric under-utilization due to load imbalance, lack of reactivity to traffic changes and component/link failures, inefficient congestion management, and inadequate admission control, leading to suboptimal throughput and latency.
Innovation Solution
The Fabric Control Protocol (FCP) enables end-to-end admission control, packet spraying across multiple paths, adaptive rate control, and explicit congestion notification, allowing for efficient bandwidth utilization, fault tolerance, and low latency by using a request-grant mechanism and packet sequence numbering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If packets are forwarded along a single path through the switch fabric, then packet ordering is maintained, but bandwidth utilization is reduced and load imbalance occurs
Solution Approach 1:
The patent segments packet flows into multiple sub-flows that can be distributed across different paths in the switch fabric. Each sub-flow is routed independently, allowing parallel transmission through multiple paths while maintaining overall flow coherence through sequence numbering and reassembly at the destination.
Solution Approach 2:
The patent introduces a new dimension of path selection by implementing multi-path forwarding capabilities. Instead of single-dimensional path selection, the system utilizes multiple dimensions of routing paths through the switch fabric, enabling packets to traverse different physical or logical paths simultaneously while maintaining delivery guarantees.
2Productivity
If multiple paths are used for packet forwarding, then bandwidth utilization improves, but packet reordering occurs
Solution Approach 1:
The patent implements feedback mechanisms through sequence numbering and acknowledgment systems. Each packet is tagged with sequence information that allows the destination to detect and correct reordering. The system uses feedback loops to track packet delivery status and ensure in-order reconstruction of the original data stream despite multi-path transmission.
Solution Approach 2:
The patent changes the parameter of packet identification by incorporating sequence numbers and path identifiers into packet headers. This parameter modification enables the destination to distinguish packets from different paths and reassemble them in the correct order, transforming the packet handling process from simple forwarding to intelligent reordering.
3Reliability
If admission control is implemented, then congestion management improves, but protocol complexity increases
Solution Approach 1:
The patent implements preliminary admission control actions before packet transmission begins. The system pre-establishes flow permissions, bandwidth allocations, and path selections through reservation protocols. This preliminary action prevents congestion before it occurs rather than reacting to it, reducing the need for complex real-time congestion management mechanisms.
Solution Approach 2:
The patent introduces intermediary control mechanisms that mediate between packet sources and the switch fabric. These intermediaries manage admission control, flow scheduling, and resource allocation, shielding the core switching fabric from complex congestion management tasks and simplifying the overall system architecture.
4Productivity
If flow-based switching is used, then packet ordering is maintained, but fabric under-utilization occurs due to load imbalance
Solution Approach 1:
The patent implements dynamic flow management that adapts to changing network conditions. Flow paths, bandwidth allocations, and routing decisions are adjusted in real-time based on fabric utilization, traffic patterns, and congestion indicators. This dynamic approach allows the system to optimize fabric utilization while maintaining flow integrity, transitioning from static flow-based switching to adaptive multi-path forwarding.
Data Source
AI summary
A fabric control protocol is described for use within a data center in which a switch fabric provides full mesh interconnectivity such that any of the servers may communicate packet data for a given packet flow to any other of the servers using any of a number of parallel data paths within the data center switch fabric. The fabric control protocol enables spraying of individual packets for a given packet flow across some or all of the multiple parallel data paths in the data center switch fabric and, optionally, reordering of the packets for delivery to the destination. The fabric control protocol may provide end-to-end bandwidth scaling and flow fairness within a single tunnel based on endpoint-controlled requests and grants for flows. In some examples, the fabric control protocol packet structure is carried over an underlying protocol, such as the User Datagram Protocol (UDP).