Host-Tagged Key-Value Deletion for Shared Storage Namespaces
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data storage systems face challenges in managing deletion of data in namespaces shared among multiple hosts, leading to conflicts and accidental deletions, particularly in disaggregated storage architectures where multiple hosts can access the same namespace.
Innovation Solution
Data storage devices append a host identifier to each value and manage a delete list to track foreign host deletions, using a deleted data structure to order and invalidate erase blocks during garbage collection, ensuring reliable deletion and recovery of key-value pairs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple hosts are granted read/write access to the same namespace, then shared access and data availability are improved, but conflicts and accidental deletions of data occur
Solution Approach 1:
The patent segments the value data by appending host identifiers to distinguish data belonging to different hosts. The value is divided into a data portion and a host identifier portion, enabling the system to track which host originally wrote the data and prevent other hosts from accidentally deleting it. This segmentation resolves the contradiction by allowing multiple hosts to access the same namespace while maintaining data integrity through host-specific identification.
Solution Approach 2:
The host identifier acts as an intermediary element between the data and the deletion operation. When a host issues a delete command, the system checks the host identifier in the value against the issuing host's identity. This intermediary mechanism mediates the interaction between multiple hosts and the shared namespace, preventing unauthorized deletions while allowing legitimate access. The host identifier serves as a mediator that enables shared access without compromising data integrity.
2Speed
If a delete command is processed immediately, then deletion speed is improved, but data recovery capability is reduced
Solution Approach 1:
The patent implements preliminary action by immediately moving the key to a deleted data structure upon receiving a delete command, while preserving the ability to recover the data. The key is marked as deleted and placed in a managed state where it can be recovered if needed. This preliminary action ensures fast deletion response while maintaining recovery capability through the preserved key information in the deleted data structure, resolving the contradiction between deletion speed and data recovery.
3Measurement precision
If host identifiers are appended to all values, then data tracking accuracy is improved, but storage overhead increases
Solution Approach 1:
The patent applies local quality by appending host identifiers only to values that require host-specific tracking, rather than uniformly to all data. The host identifier is added as a small metadata field (e.g., 1-4 bytes) only when needed for multi-host scenarios. This localized approach improves data tracking accuracy for critical data while minimizing storage overhead by avoiding unnecessary identifiers in single-host contexts. The host identifier is appended to the value as a small field, providing precise tracking with minimal impact on storage capacity.
Data Source
AI summary
Systems, methods, and data storage devices for management of key-value delete operations for namespaces supporting multiple hosts are described. Responsive to a delete command, the key for the deleted key-value pair may be moved to a deleted data structure having a delete order. During garbage collection erase blocks corresponding to the key-value pairs may be invalidated based on the delete order and removed from the deleted data structure.