Live Hardware Component Swapping in Disaggregated Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current cloud computing infrastructure is inflexible and requires downtime for hardware upgrades, leading to poor resource utilization and increased costs due to fixed computing configurations and limitations in scaling and descaling demands.
Innovation Solution
A disaggregated computing system utilizing point-to-point circuit wire level switching to dynamically swap hardware components between pools without interrupting workloads, allowing for seamless live upgrades and evaluation of new components without redundancy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional cloud computing infrastructure is used with fixed hardware configurations, then system stability is maintained, but hardware upgrades require downtime and resource utilization is poor
Solution Approach 1:
The system segments hardware resources into independent pools (compute, storage, memory, network) that can be individually upgraded. Each pool operates independently, allowing upgrades to one pool without affecting others, thus enabling hardware upgrades without system downtime.
Solution Approach 2:
The system implements dynamic reconfiguration of hardware components through point-to-point circuit wire level switching. This allows the system to dynamically swap components between pools while maintaining operational workloads, transforming static hardware configurations into dynamically adaptable ones.
2Adaptability or versatility
If hardware components are swapped using traditional methods, then new components can be installed, but workload interruption occurs
Solution Approach 1:
The system performs preliminary actions by maintaining component pools with standby components ready for immediate substitution. Before a swap is needed, alternative components are pre-positioned in the pools, enabling seamless replacement without interrupting workloads.
Solution Approach 2:
The point-to-point circuit wire level switching fabric acts as an intermediary that enables component substitution without direct interruption to workloads. The switching fabric mediates between the workload and the hardware components, allowing components to be swapped in the background while workloads continue uninterrupted.
3Reliability
If redundant hardware is maintained for failover capability, then system reliability is improved, but resource utilization decreases and costs increase
Solution Approach 1:
The component pools serve multiple functions: they act as both active hardware resources for current workloads and as standby resources for failover and upgrades. The same pools provide redundancy, upgrade capability, and resource allocation flexibility, eliminating the need for separate redundant hardware.
Solution Approach 2:
The system recovers and reuses hardware components through the pooling mechanism. When components are upgraded or fail, they are temporarily removed from service, tested or replaced, then recovered and returned to the pools for continued use, maximizing resource utilization while maintaining reliability.
Data Source
AI summary
Various embodiments for performing hardware upgrades in a disaggregated computing environment. A workload is run on a disaggregated computing system while providing a new component to at least one of a plurality of component pools used by the disaggregated computing system. Point-to-point circuit wire level switching is used to switch the disaggregated system from an assigned component residing in a first of the plurality of component pools to the new component residing in a second of the plurality of component pools without interrupting the running workload.