NVMe Target Discovery Tags for Resilient Path Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional approaches in software-defined storage systems, particularly when using advanced protocols like NVMe-oF and NVMe/TCP, face challenges in optimal target discovery, leading to sub-optimal resiliency and load balancing, which adversely impact storage system performance.
Innovation Solution
Implementing target discovery extension techniques using log page storage tags to facilitate host-side path selection, ensuring IO operations are delivered through paths with different storage tags, thereby enhancing resiliency and load balancing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional target discovery approaches are used in NVMe-oF and NVMe/TCP protocols, then the storage system can be implemented, but the resiliency and load balancing are sub-optimal
Solution Approach 1:
The patent applies preliminary action by extracting storage tags from discovery log pages before IO operations are initiated. The host device obtains target information including storage tags during the discovery phase, allowing it to pre-determine optimal paths and avoid selecting multiple paths to targets with the same storage tag. This preliminary extraction and analysis of storage tags enables the host to make informed path selection decisions before actual data operations begin, improving both resiliency and load balancing.
2Productivity
If conventional target discovery approaches are used in NVMe-oF and NVMe/TCP protocols, then the storage system can be implemented, but the load balancing is sub-optimal
Solution Approach 1:
The patent implements feedback by using storage tags extracted from discovery log pages to guide path selection decisions. The host device receives target information containing storage tags during discovery, analyzes these tags to identify paths leading to targets with different storage tags, and uses this information to make load-balanced path selection. This feedback mechanism allows the host to continuously optimize path selection based on the distributed nature of storage tags across multiple targets.
3Reliability
If paths are selected without considering storage tags, then path selection is simpler, but resiliency deteriorates due to sub-optimal path arrangements
Solution Approach 1:
The patent applies preliminary action by extracting storage tags from discovery log pages before IO operations are initiated. The host device obtains target information including storage tags during the discovery phase, allowing it to pre-determine optimal paths and avoid selecting multiple paths to targets with the same storage tag. This preliminary extraction and analysis of storage tags enables the host to make informed path selection decisions before actual data operations begin, improving both resiliency and load balancing.
Data Source
AI summary
An apparatus comprises at least one processing device that includes a processor coupled to a memory. The processing device is configured to receive in a host device, for a plurality of targets of a storage system, corresponding discovery log pages, to extract from the received discovery log pages respective storage tags for respective ones of the targets, and to control path selection for delivery of input-output operations from the host device to the targets based at least in part on the extracted storage tags. The targets illustratively comprise NVMe targets configured to store data of a plurality of logical storage volumes comprising respective NVMe namespaces of the storage system. In some embodiments, each of the storage tags indicates a failure domain of the corresponding target, such as a particular storage node, physical server or subnetwork of a storage system.