Hierarchical Tenant Sharding for Parallel Access and Isolation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multi-tenant systems face challenges in organizing and accessing data stored in hierarchical structures, leading to performance issues and security isolation, especially when dealing with large numbers of tenants and parallel data requests, where one tenant's overload can affect others.
Innovation Solution
Implementing optimized sharding techniques that allocate unique tenant data on separate shards, using hierarchical topology and bin-packing algorithms to ensure efficient distribution and minimize the number of shards required, allowing for seamless data migration and access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data is stored in a single storage solution to maintain high consistency, then data consistency is improved, but system performance and security isolation deteriorate
Solution Approach 1:
The patent divides the storage system into multiple independent shards, where each shard stores data for specific tenants. This segmentation allows parallel processing of data requests across different shards, improving system performance while maintaining data consistency within each shard. The hierarchical sharding structure organizes data into parent shards and child shards, enabling efficient data access and isolation.
Solution Approach 2:
The patent introduces a sharding management system that acts as an intermediary between the application layer and the distributed storage shards. This intermediary manages data routing, shard allocation, and hierarchical relationships, ensuring data consistency across shards while enabling parallel processing and maintaining security isolation between tenants.
2Productivity
If data is horizontally scaled by partitioning into multiple storages, then system performance is improved, but data organization and access efficiency deteriorate
Solution Approach 1:
The patent implements a hierarchical sharding structure where parent shards contain references to child shards, creating a nested organization of data. This nested structure allows the system to maintain horizontal scaling for performance while providing efficient data access through hierarchical navigation. The nested doll pattern enables organizing large distributed datasets into manageable hierarchical groups, improving both access efficiency and operational ease.
3Adaptability or versatility
If multiple tenants share common shards, then resource utilization is improved, but security isolation and system stability deteriorate
Solution Approach 1:
The patent applies local quality by assigning different isolation levels to different tenants and data types. Some tenants may share parent shards for resource efficiency, while maintaining exclusive access to specific child shards for security-critical data. This localized differentiation allows the system to optimize resource utilization while maintaining security isolation where required, preventing one tenant from overloading the system and affecting others.
Data Source
AI summary
In a multi-tenant hierarchical data storage system, tenant nodes are organized into trees and subtrees including virtual shards and with tenant data on single shards. The system is configured to allow scalable parallel access by a plurality of tenant-users.