Endpoint I/O Source ID Tracking for Virtual Storage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional storage management approaches struggle to maintain accuracy and reliability, especially in virtualized environments where virtual machines share physical components, leading to issues with data corruption protection and complex reservation management when links fail or change.
Innovation Solution
Implementing an endpoint I/O source ID value tracking process that allows unique identification and authorization of virtual machines for storage access, independent of physical connections, enabling effective I/O fencing and split link scenario handling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional SCSI-3 based I/O fencing is used, then data corruption protection is achieved, but it cannot handle link failures or changes in virtualized environments
Solution Approach 1:
The patent introduces an intermediary mechanism (the authority tracking process and endpoint I/O source ID values) that mediates between the storage target and virtual machines. This intermediary layer tracks which virtual machines have authority to access storage, independent of physical link status, thereby resolving the contradiction between maintaining data protection and adapting to link failures in virtualized environments
2Reliability
If NPIV is used to provide unique virtual adapters, then unique initiator IDs are granted to each virtual machine, but the number of virtual adapters is restricted by physical adapter capacity
Solution Approach 1:
The patent segments the unique identification mechanism from the physical adapter. Instead of relying on physical adapter limits, it creates virtual endpoint I/O source ID values that can be independently assigned to virtual machines. This segmentation allows unlimited virtual machines to have unique identifiers without being constrained by physical hardware capacity
Solution Approach 2:
The patent creates virtual copies of identification mechanisms. Rather than using physical adapter IDs that are limited by hardware, it generates virtual endpoint I/O source ID values that can be copied and assigned to any number of virtual machines, effectively unlimited by physical constraints
3Adaptability or versatility
If multiple adapters are registered with storage device, then access flexibility is improved, but reservation management becomes complicated and cumbersome
Solution Approach 1:
The patent extracts the authority tracking function from the complex reservation management process. By separating the tracking of which virtual machines have authority (endpoint I/O source ID values) from the actual reservation operations, it simplifies management while maintaining flexibility for multiple adapters and access paths
4Adaptability or versatility
If endpoint I/O source ID values are bound to virtual machines, then access authority is maintained independently of physical connections, but traditional link-dependent ID approaches fail when links change
Solution Approach 1:
The patent implements a dynamic authority tracking system that adapts to changing virtual machine positions and connections. The endpoint I/O source ID values are dynamically assigned and tracked, allowing virtual machines to maintain access authority regardless of physical link changes, thereby achieving both adaptability and reliability
Data Source
AI summary
Shared storage access management systems and methods are presented. A method can comprise: performing an endpoint I/O source authority tracking process in which permission of an endpoint I/O source to perform an I/O with at least a portion of an endpoint storage target is tracked at least in part based on a corresponding endpoint I/O source ID value, and performing an I/O operation based upon results of the endpoint I/O source authority tracking process. In one embodiment, the endpoint I/O source ID value is associated with an endpoint I/O source that is a virtual machine. The endpoint I/O source ID value can be unique and can be bound to an endpoint I/O source. In one exemplary implementation, the endpoint I/O source ID value does not depend upon intermediate communication channel characteristics between a corresponding endpoint I/O source and endpoint storage target.