Fabric Storage Multicast with Distributed Intelligence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current fabric storage systems with 'intelligent' switches slow down request delivery due to the need for switch intelligence to process and generate tailored requests for storage devices, especially when data is striped across multiple devices.
Innovation Solution
Distributing intelligence to storage devices through multicast functionality, allowing them to determine their storage layout and respond independently to requests, thereby reducing switch intelligence and speeding up request delivery.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If an intelligent switch with mapping functionality is used to generate tailored storage requests, then storage request routing accuracy is improved, but request delivery speed deteriorates
Solution Approach 1:
The mapping functionality is extracted from the switch and relocated to the storage devices themselves. Each storage device now contains the mapping information needed to determine its storage layout, eliminating the need for the switch to process and generate tailored requests, thus resolving the contradiction between routing accuracy and delivery speed
Solution Approach 2:
Storage devices are empowered with self-service capability through embedded mapping functionality. Each device can independently determine whether it should respond to a multicast request based on its own storage layout information, eliminating the need for switch-based request tailoring while maintaining accurate routing
2Adaptability or versatility
If mapping functionality is concentrated in the switch, then centralized control is improved, but system complexity increases
Solution Approach 1:
The centralized mapping functionality is segmented and distributed to individual storage devices. Instead of one switch containing all mapping information, each storage device holds its own mapping data, reducing switch complexity while maintaining control through the multicast mechanism
Solution Approach 2:
The switch's role is simplified to universal multicast distribution, while storage devices gain multi-functionality by combining data storage with mapping functionality. This redistribution of functions reduces overall system complexity, particularly at the switch level
Data Source
AI summary
The present invention is a system utilizing multicast with distributed intelligence including an initiator device for transmitting a request, the request being addressed to a multicast group. The system also includes a switch for receiving the request from the initiator device. The switch is configured with multicast functionality for multicasting copies of the request to the multicast group. Additionally, the system includes a plurality of multicast group devices, each configured for receiving a copy of the request from the initiator device, via the switch. Each device of the plurality of multicast group devices is further configured with mapping functionality for allowing the storage device to determine a storage layout of the multicast group device. The initiator device, switch and each device of the plurality of multicast group devices are communicatively coupled via a network.