Minimally Degraded Server Configuration via Associative Group Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for deconfiguring server systems with failing connections often result in excessive processor deconfigurations, leading to reduced system performance due to the need to deconfigure multiple processors when a single failure occurs, especially in systems with tightly coupled processors.
Innovation Solution
A method and computer program product that determine a minimally degraded configuration by identifying associative deconfigurations, grouping hardware items, and selectively deconfiguring endpoints of failed connections based on the smallest failed connection count to minimize system downtime and maintain performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current deconfiguration methods are used when connection failures occur, then connection failures are handled, but excessive processors are deconfigured leading to reduced system performance
Solution Approach 1:
The patent segments the system into associative groups of hardware items based on their interconnection relationships. By dividing the system into these logical groups, the deconfiguration process can be localized to specific groups rather than affecting the entire system, thus resolving the contradiction between handling connection failures reliably and maintaining overall system performance.
Solution Approach 2:
The patent applies local quality by determining deconfiguration needs at the level of individual associative groups rather than uniformly across the entire system. Each group's deconfiguration status is independently determined based on its specific connection failures, allowing healthy groups to continue operating at full performance while only affected groups are deconfigured.
2Reliability
If deconfiguration of hardware items is performed at both endpoints of failed connections, then connection failures are addressed, but system performance is greatly reduced potentially leaving only one drawer or processor functional
Solution Approach 1:
The patent introduces dynamic state tracking for associative groups (unknown, deconfigured, configured) and uses iterative processing to adaptively determine deconfiguration needs. This dynamic approach allows the system to start with conservative assumptions and progressively deconfigure only when necessary, rather than applying static blanket deconfiguration rules that would overly reduce system capacity.
Solution Approach 2:
The patent implements feedback mechanisms by counting failed connections within and between associative groups, and using this information to iteratively update deconfiguration decisions. The system continuously monitors the impact of deconfiguration decisions and adjusts subsequent decisions based on this feedback, ensuring that deconfiguration is minimized while still addressing connection failures reliably.
3Ease of manufacture
If associative deconfigurations are applied when deconfiguring processors on a drawer, then hardware simplicity is maintained, but unnecessary processors are deconfigured reducing system performance
Solution Approach 1:
The patent applies partial action by selectively applying associative deconfiguration only when necessary, rather than always deconfiguring entire drawers. The system determines the minimal set of hardware items that need to be deconfigured based on actual connection failures, avoiding excessive deconfiguration while still maintaining hardware simplicity through the associative group framework.
Data Source
AI summary
A minimally degraded configuration is determined when failing connections occur. Associative deconfigurations are determined from deconfiguring hardware items in a server system, associative groups are derived, and failed connections are determined. Failed connections are determined between two hardware items that are in the same associative group, and the two hardware items at both endpoints of the failed connection are deconfigured. Each associative group state is set to unknown, and the failed connections are counted where a single endpoint of the failed connection is within the associative group. The associative group state is set to deconfigured, if a member of the associative group was deconfigured. Counts of the associative groups that remain in the unknown state are analyzed, and the associative group with the smallest failed connection count is selected and set to a configured state. The other associative group at the other end is set to a deconfigured stated, and the hardware item in the other associative group is deconfigured.