Coordinating Unplanned Storage Swaps Across Clusters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sysplex environments lack the ability to coordinate unplanned swap events across multiple computing clusters effectively, leading to potential data inconsistencies and system disruptions due to independent swap functions and lack of communication protocols.
Innovation Solution
The method involves receiving notifications from each computing cluster through a shared storage device, determining the capable target storage devices for swapping, and coordinating the swap event by using attention interrupts and Set Sysplex Replication Status messages to ensure all clusters agree on the target storage devices. If agreement is not reached, shared storage devices are disabled to maintain data consistency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If each computing cluster performs independent swap functions without coordination, then each cluster can swap to its preferred target storage device, but data inconsistencies and system disruptions occur across multiple clusters
Solution Approach 1:
A coordination mechanism is introduced as an intermediary between multiple computing clusters to manage swap events. The system uses shared storage devices as mediators that can hold notifications from multiple clusters and coordinate the swap process, ensuring that all clusters agree on the target storage device before executing the swap, thereby preventing data inconsistencies while maintaining swap capability.
2Reliability
If a coordination mechanism is introduced across multiple computing clusters, then data consistency is maintained during swap events, but system complexity increases due to communication protocols and coordination overhead
Solution Approach 1:
The coordination mechanism leverages existing shared storage devices that are already part of the computing cluster infrastructure. These storage devices self-serve as notification mediums and coordination points without requiring external coordination systems. The clusters communicate swap intentions and capabilities through the shared storage devices using standard I/O operations, reducing the need for additional complex coordination infrastructure.
3Reliability
If shared storage devices are disabled to maintain data consistency, then system reliability is improved during coordination failures, but storage device availability and productivity decrease
Solution Approach 1:
Instead of disabling all shared storage devices when coordination issues arise, the system applies partial action by selectively disabling only the specific storage devices involved in the conflicting swap event while leaving other storage devices operational. This approach maintains data consistency for the affected devices while preserving productivity and availability of the broader storage system.
Data Source
AI summary
Coordinating an unplanned swap event across multiple computing clusters includes receiving a notification from each of a plurality of computing clusters through a shared storage device of a set of storage devices. The notification identifies which of a first set of target storage devices and a second set of target storage devices that a particular computing cluster of the plurality of computing clusters is capable of swapping to from a set of source storage devices. Based on the notification from each of the plurality of computing clusters, it is determined whether the plurality of computing clusters will swap from usage of the set of source storage devices to the first set of target storage devices or to the second set of target storage devices during a swap event.