Mixed Distributed Database Data Partitioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current distributed database systems face limitations in scalability and availability, as they are either restricted by shared-nothing or shared-disk architectures, which do not effectively manage data and workload in real application environments.
Innovation Solution
A mixed distributed database system is proposed, combining shared-nothing and shared-disk architectures, where nodes are independent but can share disks within a node, allowing for data partitioning and dynamic load balancing, with a data manager that divides data into partitions and stores them across multiple nodes, including sharing data groups and single nodes, and uses a Paxos-Plus method for leader node selection in case of failures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a distributed database system uses shared-nothing architecture, then system scalability is improved, but data sharing capability deteriorates
Solution Approach 1:
The patent combines shared-nothing and shared-disk architectures into a hybrid model where data can be partitioned across nodes (shared-nothing characteristic) while also allowing controlled sharing of data partitions among multiple databases within nodes (shared-disk characteristic). This merging resolves the contradiction by enabling both scalability through partitioning and data sharing through the hybrid architecture.
Solution Approach 2:
The system segments data into partitions that can be distributed across multiple nodes for scalability, while also allowing flexible assignment of these partitions to different databases. This segmentation enables the system to maintain both scale and sharing capability by organizing data in manageable units that can be selectively shared.
2Device complexity
If a distributed database system uses shared-disk architecture, then data sharing capability is improved, but system scalability deteriorates
Solution Approach 1:
The patent segments the shared-disk architecture into manageable data partitions that can be selectively distributed. Instead of sharing entire disks across all nodes, the system divides data into partitions that can be shared only where needed, enabling scalability while maintaining data sharing capability through controlled partition distribution.
Solution Approach 2:
The system applies local quality by allowing different nodes to have different data partitions based on their specific needs. Each node can access and share only the partitions relevant to its databases, rather than requiring all nodes to share all data, thus improving scalability while maintaining necessary data sharing capability.
3Reliability
If data is partitioned and stored across multiple nodes, then system availability is improved, but system complexity deteriorates
Solution Approach 1:
The patent implements self-service through automated partition management and failover mechanisms. When nodes fail or are added, the system automatically redistributes partitions and updates database assignments without manual intervention, reducing the operational complexity of managing distributed data while maintaining high availability through automatic fault tolerance.
Data Source
AI summary
Embodiments of the present disclosure relate to a method, system, and computer program product for data management. According to the method, one or more processors divide data into a plurality of partitions. The one or more processors store the plurality of partitions in a plurality of nodes of a mixed distributed database system, wherein a first node of the mixed distributed database system comprises a plurality of databases, and wherein at least a part of the plurality of partitions are shared by the plurality of databases of the first node and being not shared by other of the plurality of nodes.