Metadata Routing Coordinator for Distributed Database Scalability

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In multi-partitioned databases, clients face difficulties in determining the responsible metadata node, leading to network congestion and scalability limitations due to excessive network requests for node discovery.

Innovation Solution

A distributed system that maintains strongly consistent metadata copies across multiple nodes using the Raft consensus algorithm, with each partition having a single leader node and client-side caching for efficient request routing, allowing clients to directly communicate with the correct node and fallback mechanisms to ensure high availability and scalability.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If metadata is replicated across multiple nodes for high availability, then system reliability is improved, but clients face difficulty determining which node is responsible for the metadata they need, leading to excessive network requests

Engineering Contradiction:
Improvehigh availability of metadataVSAvoiddifficulty in determining responsible node
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent introduces a coordinator node as an intermediary that maintains a mapping between metadata keys and responsible metadata nodes. Clients query the coordinator to obtain the correct node for their metadata access needs, eliminating the need for clients to directly determine node responsibility and reducing unnecessary network requests across the distributed system.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Loss of information

If clients send requests to all metadata nodes to discover the correct node, then the correct node can be located, but network congestion occurs and system scalability is limited

Engineering Contradiction:
Improvenode discovery accuracyVSAvoidsystem scalability
Core Design Contradiction:
Loss of informationVSProductivity

Solution Approach 1:

The patent extracts the node discovery function from the general metadata access process by creating a specialized coordinator node that handles location queries. This separates the routing function from data access, allowing clients to efficiently locate the correct metadata node through the coordinator without broadcasting requests to all nodes, thereby maintaining node discovery accuracy while enabling system scalability.

Inventive Principle:
Principle #2Taking out (Extraction)

3Ease of operation

If a centralized approach is used for metadata management, then node routing is simplified, but system flexibility for reconfiguration and dynamic scaling is reduced

Engineering Contradiction:
Improvenode routing simplicityVSAvoidsystem reconfiguration flexibility
Core Design Contradiction:
Ease of operationVSAdaptability or versatility

Solution Approach 1:

The patent implements dynamic metadata partitioning where the coordinator node maintains flexible mappings that can be reconfigured based on system state. Metadata can be dynamically assigned to different nodes, and the coordinator updates its mapping accordingly. This allows the system to adapt to changing conditions, scale dynamically, and reconfigure partitions without requiring a fixed centralized structure, thereby maintaining routing simplicity while enabling system flexibility.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3935515B1Metadata routing in a distributed system
Publication Date: 2024.02.14 HITACHI VANTARA LLC
  • EP3935515B1 patent drawingFigure 1
  • EP3935515B1 patent drawingFigure 2
  • EP3935515B1 patent drawingFigure 3

AI summary

In some examples, a first computing device may receive, from a second computing device, partition mapping information indicating partitions of a metadata database. The first computing device may be able to communicate with a plurality of metadata nodes, each metadata node maintaining a portion of the metadata database based on the partitioning of the metadata database to distribute the metadata database across the plurality of metadata nodes. The first computing device may determine to send a request to the metadata database based at least on key information. The first computing device may determine, based on the partition mapping information, a first metadata node of the plurality of metadata nodes indicated to maintain a partition of the metadata database corresponding to the key information. The first computing device, may send, to the first metadata node, based on the partition mapping information, a request to perform a database operation.