Host-Specific Protocol Endpoints for Virtual Volume Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In data storage systems, existing technologies face challenges in minimizing latency for virtual volumes bound to protocol endpoints, balancing IO load, scaling SCSI command handling, and managing out-of-band operations efficiently, especially in environments with multiple ESXi hosts accessing the same set of protocol endpoints.
Innovation Solution
The method involves organizing physical storage into virtual volumes and creating protocol endpoints, which are grouped into host-specific initiator groups, with each group having a pair of endpoints advertised in active-optimized and active-unoptimized states on different processing nodes, allowing for shared processing and failover, and limiting Unit Attention messages to relevant hosts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If virtual volumes are bound to protocol endpoints in a clustered storage system, then data access is enabled for virtual machines, but latency increases due to distributed processing across multiple nodes
Solution Approach 1:
The patent segments the protocol endpoint functionality by creating host-specific initiator groups that separate VVols bound to the same host into dedicated groups. This segmentation allows each host to access only its relevant VVols through its specific IG, avoiding the need to process requests for all VVols across the cluster, thereby reducing latency while maintaining data access capability.
2Productivity
If protocol endpoints serve multiple virtual volumes for multiple hosts, then resource utilization improves, but IO load balancing becomes difficult across storage appliances and processing nodes
Solution Approach 1:
The patent applies local quality by creating host-specific initiator groups with localized management. Each IG is specifically tailored to a particular host's requirements and contains only the VVols relevant to that host. This localized approach enables precise control over IO load distribution, as each host's requests are directed to its dedicated IG, making load balancing straightforward while maintaining high resource utilization across the cluster.
3Measurement precision
If Unit Attention messages are raised on a per VVol basis for SCSI command handling, then data state tracking is improved, but message scaling becomes problematic when many hosts access the same protocol endpoints
Solution Approach 1:
The patent merges the tracking of data state and path state into a unified host-specific initiator group framework. By combining VVol-specific data state tracking with host-specific path state management in the same IG structure, the system achieves precise data state tracking while simplifying message scaling. Unit Attention messages can be efficiently targeted to specific hosts through their dedicated IGs, reducing the complexity of managing messages across multiple hosts accessing the same PEs.
4Loss of information
If out-of-band management reports all protocol endpoints to ESXi hosts, then complete visibility is provided, but management scaling becomes inefficient with irrelevant PEs reported
Solution Approach 1:
The patent extracts and isolates only the relevant protocol endpoints for each host by creating host-specific initiator groups. Instead of reporting all PEs to every ESXi host, the system extracts and reports only those PEs that are actually bound to VVols accessible by each specific host. This extraction approach maintains complete visibility of relevant resources while dramatically improving management scaling efficiency by eliminating irrelevant PE reports.
Data Source
AI summary
Data storage systems of a cluster associate virtual volumes (VVols) with protocol endpoints (PEs) organized into host-specific initiator groups. From the perspective of the hosts, the PEs are specialized SCSI LUNs that aggregate the VVols bound thereto. In operation, a host directs VVol access requests to the host-specific PEs, and a data storage system processes the VVol requests as received via the corresponding PEs.