Affinity-Based Access Control for Distributed Storage Consistency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Distributed data storage systems face challenges in maintaining data consistency during communication losses between sites, leading to potential inconsistencies and inhibited access for clients, as existing solutions often restrict access to only one site during communication breakdowns.
Innovation Solution
The system employs affinity-based access privileges to allow multiple sites to access data during communication loss and automatically resolves differences upon restoration, using affinity mappings and plug-in resolvers to manage IO operations and maintain data consistency across sites.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If bias or voting rules are used to limit access to the LUN during communication breakage, then data consistency is maintained, but client progress at sites without access is inhibited
Solution Approach 1:
The data set is divided into multiple portions or ranges, with each portion assigned to a specific site through affinity mapping. During communication breakage, each site can independently access its assigned portions without requiring coordination with other sites, thereby maintaining data consistency while allowing client progress at all sites.
Solution Approach 2:
Different sites are granted different access permissions to different portions of the data set based on their affinity. Site 1 may have read-write access to portion A while Site 2 has read-write access to portion B, allowing each site to operate independently on its local portion without affecting others.
2Reliability
If only one site is permitted to access the LUN during communication loss, then data consistency is ensured, but access availability at other sites is reduced
Solution Approach 1:
The data set is segmented into multiple portions, each assigned to different sites through affinity mapping. This segmentation allows multiple sites to simultaneously access different portions of the data set during communication loss, maintaining consistency while improving access availability.
Solution Approach 2:
The system transitions from a single-dimension access model (one site accesses entire LUN) to a multi-dimensional model where different sites access different portions of the LUN simultaneously. This dimensional change enables parallel access across multiple sites without compromising consistency.
Data Source
AI summary
In a distributed storage system in which a first copy of a set of data such as a virtualized LUN is maintained by a first site, and a second copy is maintained by a second site, access is provided to both the first site and the second site during loss of communication between the first site and the second site. Affinity determines access privileges for each site, where affinity is an indication that a particular site has or should be given permission to perform an operation to a particular part of the data set. Affinities can be specified such that permissions are non-overlapping, or overlap in a manner which is acceptable and resolvable.