Data-Consistent Freeze Coordination for Peer-to-Peer Remote Copy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for performing consistent freeze across multiple computing clusters in a multi-cluster environment, such as z/OS HyperSwap, face challenges in minimizing User Impact Time (UIT) due to unpredictable delays and potential data inconsistency caused by asynchronous thawing operations.
Innovation Solution
A method is introduced where each computing cluster freezes non-shared logical volumes first, followed by shared volumes, with a designated key logical volume determined by a numerically minimum identifier, ensuring a shared understanding of which cluster thaws the shared volumes last, thereby minimizing UIT.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If each computing cluster independently performs thaw operations on shared logical volumes, then operational independence is maintained, but data inconsistency occurs and User Impact Time increases
Solution Approach 1:
The patent introduces a coordinator computing cluster that acts as an intermediary to manage thaw operations across multiple computing clusters. The coordinator receives freeze requests, determines which cluster should perform the thaw operation, and communicates this designation to other clusters. This mediator approach maintains operational independence of individual clusters while ensuring centralized coordination to prevent data inconsistency and reduce User Impact Time.
Solution Approach 2:
The system implements feedback mechanisms where computing clusters report their freeze status and path group disbandment completion to the coordinator. The coordinator uses this feedback information to determine when all clusters have completed their freeze operations and which cluster should perform the thaw. This feedback loop ensures data consistency while maintaining the ability of individual clusters to operate independently.
2Reliability
If computing clusters wait for all other clusters to complete freeze operations before thawing, then data consistency is maintained, but User Impact Time increases due to unpredictable delays
Solution Approach 1:
The coordinator computing cluster performs preliminary actions by receiving freeze requests from all computing clusters before any thaw operations begin. It determines in advance which cluster should perform the thaw operation based on the coordinator's identity and the freeze completion status. This preliminary determination eliminates unpredictable waiting delays during the thaw phase while maintaining data consistency through centralized coordination.
Solution Approach 2:
The patent segments the freeze and thaw operations into distinct phases with clear coordination points. The freeze phase allows parallel execution across all clusters, and the thaw phase is coordinated centrally. This segmentation enables predictable timing and reduces User Impact Time by eliminating the need for clusters to wait for each other's freeze completion before initiating their own thaw operations.
3Reliability
If a coordinator computing cluster is introduced to manage freeze and thaw operations, then data consistency and UIT optimization are achieved, but system complexity increases
Solution Approach 1:
The coordinator computing cluster performs multiple functions: it receives freeze requests, determines thaw operation assignment, tracks freeze completion status, and communicates with all computing clusters. By consolidating these coordination functions into a single multi-functional entity, the patent avoids the need for complex peer-to-peer coordination protocols between multiple clusters, thereby managing system complexity while achieving data consistency and UIT optimization.
4Adaptability or versatility
If computing clusters disband path groups asynchronously, then operational flexibility is maintained, but data inconsistency occurs
Solution Approach 1:
Computing clusters provide feedback to the coordinator when they have completed disbanding their path groups for the shared logical volumes. The coordinator uses this feedback to determine when all clusters have completed their freeze operations and to identify which cluster should perform the thaw. This feedback mechanism maintains operational flexibility for asynchronous path group disbandment while ensuring data consistency through centralized coordination of the thaw operation.
Data Source
AI summary
Data consistent freeze of peer-to-peer remote copy in a multiple computing cluster environment includes freezing, by each of a first computing cluster and a second computing cluster, non-shared logical volumes of the first computing cluster and the second computing cluster; freezing, by each of the first computing cluster and the second computing cluster, shared logical volumes of the first computing cluster and the second computing cluster, wherein a key logical volume is a last shared logical volume to be frozen; selecting one of the first computing cluster and the second computing cluster to perform thaws of the shared logical volumes based on which of the first computing cluster or the second computing cluster was the last to freeze the key logical volume; and thawing, by the selected one of the first computing cluster and the second computing cluster, the shared logical volumes.