Hierarchical Network Management for Multi-Board Data Center Chassis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The efficient organization of components in a data center computing system to deliver applications and data to users is challenging due to the complexity of managing operations across various computing devices and networks.
Innovation Solution
A data center computing system with a hierarchical management structure that includes an instruction processing device to manage operations of computing devices, utilizing two distinct communication networks with different protocols to facilitate communication between controllers and processing devices within a chassis, enabling efficient operation and management of multiple boards and cards.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a single communication network and protocol are used to manage all computing devices, then the system structure is simple, but the efficiency of communication between controllers and processing devices is insufficient
Solution Approach 1:
The communication network is segmented into two distinct networks: a first communication network connecting the instruction processing device to controllers, and a second communication network connecting processing devices to each other. This segmentation allows each network to be optimized for its specific communication needs, improving overall communication efficiency while maintaining manageable complexity through clear separation of concerns.
Solution Approach 2:
Controllers act as intermediaries between the instruction processing device and processing devices. The first communication network carries instructions from the instruction processing device to controllers, which then forward these instructions through the second communication network to the appropriate processing devices. This intermediary structure enables efficient communication by allowing specialized protocols for each network segment.
2Productivity
If multiple communication protocols are used for different networks, then the communication efficiency is improved, but the system complexity increases
Solution Approach 1:
Different communication protocols are applied locally to different network segments based on their specific requirements. The first communication network uses a protocol optimized for instruction transmission from the instruction processing device to controllers, while the second communication network uses a protocol optimized for data exchange between processing devices. This local optimization improves workload orchestration efficiency without requiring the entire system to accommodate multiple protocols simultaneously.
Solution Approach 2:
The system segments protocol usage by network function: management and control instructions use one protocol over the first network, while operational data exchange uses another protocol over the second network. This segmentation reduces protocol management complexity by confining each protocol to its appropriate domain, making the system easier to manage despite using multiple protocols overall.
3Ease of operation
If a hierarchical management structure is implemented, then the organization of computing devices is improved, but the difficulty of detecting and measuring system operations increases
Solution Approach 1:
The hierarchical management structure is segmented into distinct levels: the instruction processing device at the top level, controllers at the intermediate level, and processing devices at the operational level. Each level has clearly defined responsibilities and communication interfaces, improving system organization. The segmentation also facilitates monitoring by allowing each level to be observed and measured independently through its specific communication network.
Solution Approach 2:
Controllers serve as intermediaries that simplify monitoring of the hierarchical structure. They receive instructions from the instruction processing device through the first communication network and transmit operational data back through the same network, while also managing communication with processing devices through the second network. This intermediary role makes the hierarchical structure easier to detect and measure by providing clear observation points at each level.
Data Source
AI summary
The present disclosure describes a data center system including an instruction processing device for a chassis including one or more boards, a board including a set of cards, a card including a computing node or a processing device and a controller. The controller can receive an instruction from the instruction processing device through a first communication network between the controller and the instruction processing device. Based on the instruction, the processing device of the card can generate an operation result and provide the operation result to another device of the board through a second communication network different from the first communication network. The second communication network can have a ring topology or other topology. The instruction processing device can further operate a hierarchical manager to create a network manager to manage the operations of the board.