Ring-Configured DC Power Supplies for Compute Node Redundancy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing scalable computer systems require a large number of dedicated power supplies for redundancy, leading to increased cost and complexity, as each compute node needs at least two dedicated power supplies for fault tolerance, which is inefficient as the number of power supplies scales with the number of compute nodes.
Innovation Solution
A scalable computer system architecture with redundant power supplies arranged in a ring or hypercube configuration, where each compute node receives regulated DC voltages from multiple power supplies, reducing the total number of power supplies needed while maintaining redundancy, with each compute node connected to two or more power supplies in a ring configuration or multiple power supplies in a hypercube configuration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If each compute node is equipped with dedicated power supplies for redundancy, then fault tolerance and reliability are improved, but the total number of power supplies increases, leading to increased cost and complexity
Solution Approach 1:
Each DC power supply is designed to serve multiple compute nodes simultaneously. The power supplies are connected in a ring configuration where each power supply can provide power to two different compute nodes, allowing the same power supply to fulfill multiple functions and serve multiple purposes across the system.
Solution Approach 2:
The patent merges the power supply resources by connecting multiple compute nodes to shared DC power supplies in a ring configuration. Instead of having dedicated power supplies for each compute node, the system combines power supply resources so that N DC power supplies can support N compute nodes through shared connectivity, reducing the total number of power supplies required.
2Reliability
If the number of power supplies is increased to provide redundancy for each compute node, then reliability is improved, but cost and system complexity increase
Solution Approach 1:
Each DC power supply is designed to serve multiple compute nodes simultaneously. The power supplies are connected in a ring configuration where each power supply can provide power to two different compute nodes, allowing the same power supply to fulfill multiple functions and serve multiple purposes across the system.
Solution Approach 2:
The system provides more than the minimum required redundancy by ensuring that each compute node can draw power from multiple sources in the ring configuration, allowing partial operation to continue even when multiple power supplies fail, thus providing excessive action beyond basic redundancy requirements.
3Reliability
If dedicated power supplies are assigned to each compute node for fault tolerance, then reliability is improved, but the scaling efficiency deteriorates as the system grows
Solution Approach 1:
Each DC power supply is designed to serve multiple compute nodes simultaneously. The power supplies are connected in a ring configuration where each power supply can provide power to two different compute nodes, allowing the same power supply to fulfill multiple functions and serve multiple purposes across the system.
Solution Approach 2:
The patent transitions from a one-to-one mapping between power supplies and compute nodes to a many-to-many relationship through the ring configuration. This dimensional change in the connectivity model allows N power supplies to support N compute nodes with redundancy, breaking the traditional linear scaling constraint where power supply count must exceed compute node count for redundancy.
Data Source
AI summary
Scalable computer systems with redundant power supplies for improved fault tolerance and reliability are provided. The system includes a plurality of compute nodes, and includes a plurality of DC power supplies arranged in a ring configuration with the compute nodes. The first compute node is electrically connected to the last DC power supply. Each compute node is coupled to receive regulated DC voltages from two of the plurality of DC power supplies. The compute nodes may include processors, storage devices and/or memory devices.