Rack Scale Design for NVMeOF Resource Pooling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current enterprise/cloud computer systems face inefficiencies in resource utilization due to over-allocation of resources to meet performance requirements, leading to higher total cost of ownership (TCO) and lower return on investment (ROI), particularly in datacenter environments where resources are not optimally managed at the rack level.
Innovation Solution
The Rack Scale Design (RSD) architecture disaggregates compute, storage, and network resources, enabling their pooling and dynamic composition based on workload demands, using Pod Managers, Pooled System Management Engines, and a configurable network fabric to optimize resource allocation and management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If resources are over-allocated to meet performance requirements, then performance requirements are satisfied, but total cost of ownership increases and return on investment decreases
Solution Approach 1:
The patent implements dynamic resource allocation where compute and storage resources are dynamically assigned to racks based on actual workload demands rather than static over-allocation. The system continuously monitors performance metrics and adjusts resource distribution to meet performance requirements only when needed, reducing unnecessary resource consumption and associated costs.
Solution Approach 2:
The system changes allocation parameters by transitioning from fixed resource quotas to demand-based allocation thresholds. Performance requirements are translated into dynamic allocation parameters that adjust resource distribution in real-time, allowing the system to satisfy performance SLAs while minimizing total cost of ownership through optimized resource utilization.
2Ease of operation
If resources are not optimally managed at the rack level, then resource allocation is simplified, but resource utilization efficiency decreases
Solution Approach 1:
The patent segments resource management at the rack level by introducing rack-level controllers that independently manage compute and storage resource allocation for each rack. This segmentation enables localized optimization of resource utilization within each rack while maintaining overall system simplicity through standardized allocation policies and automated management interfaces.
Solution Approach 2:
The system implements feedback mechanisms where performance metrics collected at the rack level feed back into allocation decisions. Resource allocation is continuously adjusted based on actual performance data, enabling the system to achieve both simplified operations through automated feedback loops and optimized resource utilization efficiency through data-driven allocation.
Data Source
AI summary
Non-volatile Memory Express over Fabric (NVMeOF) using Volume Management Device (VMD) schemes and associated methods, systems and software. The schemes are implemented in a data center environment including compute resources in compute drawers and storage resources residing in pooled storage drawers that are communicatively couple via a fabric. Compute resources are composed as compute nodes or virtual machines/containers running on compute nodes to utilize remote storage devices in pooled storage drawers, while exposing the remote storage devices as local NVMe storage devices to software running on the compute nodes. This is facilitated by virtualizing the system's storage infrastructure through use of hardware-based components, firmware-based components, or a combination of hardware/firmware- and software-based components. The schemes support the use of remote NVMe storage devices using an NVMeOF protocol and/or use of non-NVMe storage devices using NVMe emulation.