Disaggregated Memory Instantiation on PCIe Fabric
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional networked storage and remote computing systems face physical limitations such as density constraints, space requirements for interconnects and climate control, and limitations on the number of devices per host, which hinder scalability and reliability in high-capacity, redundant storage environments.
Innovation Solution
The implementation of a disaggregated computing architecture using a PCIe fabric that allows for logical partitioning and dynamic reconfiguration of compute units, enabling flexible resource allocation and management through a management processor that routes traffic and manages resources across a cluster of compute units.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If networked storage systems and remote computing systems are increased in density, then storage capacity and computing resources are improved, but physical space requirements for interconnects and climate control systems increase
Solution Approach 1:
The system segments storage and computing resources into independent units that can be dynamically allocated. Storage devices are separated from host systems and organized into networked storage pools, allowing multiple hosts to share storage resources without requiring each host to have dedicated physical storage capacity. This segmentation enables efficient space utilization while maintaining high storage capacity.
Solution Approach 2:
The networked storage system provides universal access to storage resources across multiple hosts and applications. A single storage pool can serve multiple purposes including primary storage, backup, archival, and data lake functions for different hosts simultaneously. This multi-functionality maximizes the utilization of physical storage space while supporting diverse computing workloads.
2Reliability
If the number of devices per host is increased, then storage capacity and redundancy are improved, but device complexity and management difficulty increase
Solution Approach 1:
The patent introduces a storage manager and network fabric as intermediaries between hosts and storage devices. The storage manager abstracts the complexity of managing multiple storage devices by providing a unified interface for hosts to access storage resources. It handles device provisioning, monitoring, and coordination, thereby reducing the complexity burden on individual hosts while enabling high redundancy through multiple storage devices.
Solution Approach 2:
The system implements feedback mechanisms where the storage manager continuously monitors the status, health, and performance of storage devices and hosts. Based on this feedback, the system can dynamically adjust resource allocation, detect failures, and initiate recovery procedures. This feedback loop simplifies management by providing automated oversight while maintaining high reliability through continuous monitoring and adaptive response to system conditions.
Data Source
AI summary
Disaggregated computing architectures, platforms, and systems are provided herein. In one example, a method includes instructing a communication fabric to establish a first logical partition in the communication fabric that includes a first processing device and a memory device, and directing transfer of configuration data for storage by the memory device over the first logical partition. After transfer of the configuration data, the method includes instructing the communication fabric to remove the first logical partition in the communication fabric, where the configuration data remains stored by the memory device after removal of the first logical partition. The method also includes instructing the communication fabric to establish a second logical partition in the communication fabric that includes at least a second processing device and the memory device comprising the configuration data, where the second processing device is operated using the configuration data.