Virtual Volume Converter I/O Filter Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage systems struggle to efficiently manage virtual volumes (VVols) by intercepting and redirecting I/O operations without performance bottlenecks, especially when migrating legacy storage arrays that do not support VVol functionality.
Innovation Solution
A control-path virtual machine (VM) is created to manage back-end devices and map VVols, using an I/O filter to intercept and reroute I/O operations directly to data storage devices, allowing for scalable and distributed data-path handling without centralized bottlenecks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If I/O operations are intercepted and redirected through a centralized control path, then VVol management and legacy storage migration are enabled, but performance bottlenecks occur
Solution Approach 1:
The system segments the storage management architecture into control path (for management operations) and data path (for I/O operations). The I/O filter is inserted in the data path to intercept and redirect I/O operations directly to backend storage devices, separating the performance-critical data path from the management-focused control path, thereby eliminating performance bottlenecks while maintaining VVol management capabilities
Solution Approach 2:
The I/O filter acts as an intermediary component in the data path that intercepts I/O operations and redirects them to the appropriate backend storage devices. This intermediary enables legacy storage array migration to VVol paradigm without forcing all I/O through the centralized control path, thus maintaining performance while enabling adaptability
2Ease of operation
If all I/O operations pass through a centralized control path manager, then unified VVol management is achieved, but scaling is limited
Solution Approach 1:
The architecture segments management functions into control path (handled by control path manager VM) and data path operations (handled by distributed I/O filters on individual hosts). This segmentation allows unified VVol management through the control path while enabling independent scaling of data path operations across multiple hosts without centralized bottlenecks
Solution Approach 2:
The system adds a distributed data path dimension parallel to the centralized control path. I/O filters are deployed across multiple hosts in the data path, enabling scaling across the infrastructure while the control path manager maintains unified management. This dimensional addition allows both centralized management and distributed scaling to coexist
Data Source
AI summary
In one aspect, a method includes providing virtual volumes (VVols) and mappings from the VVols to corresponding data storage devices to an I/O filter in a first virtual machine (VM), sending control path commands sent from the first VM to a control-path manager VM, the first VM and the control-path manager VM being run on a VM host, intercepting an I/O for a VVol using the I/O filter and sending the intercepted I/O to a data storage device mapped to the VVol.


