Storage Volume Area Migration for Load Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage systems cannot scale out the performance of individual volumes even when additional servers are added, limiting the response time and throughput for high-access-load volumes.
Innovation Solution
A storage system where each node monitors the load of the volumes and areas within them, migrating areas to other nodes when the load exceeds a threshold, allowing for dynamic load balancing and performance scaling across multiple nodes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If data is stored in local storage of a single server, then access speed is improved, but the performance of individual volumes cannot be scaled out even when servers are added
Solution Approach 1:
The volume is divided into multiple areas that can be distributed across different servers. Each area can be independently managed and accessed, allowing the volume's performance to scale with the number of servers while maintaining fast local access for frequently accessed data segments.
Solution Approach 2:
The patent introduces a new dimension of storage architecture by allowing a single volume to span multiple servers horizontally. This enables volume performance to scale across the network dimension rather than being constrained to a single server's local storage capacity.
2Device complexity
If all volume areas are accessed via a single server with local storage, then simple architecture is maintained, but the performance becomes limited to the upper limit of that server
Solution Approach 1:
By segmenting the volume into multiple areas distributed across different servers, the system achieves higher performance without significantly increasing architectural complexity. The management system handles the distribution and access routing transparently.
Solution Approach 2:
The storage system achieves multi-functionality by enabling volumes to dynamically utilize resources from multiple servers. This allows the system to adapt to varying performance requirements while maintaining a unified storage interface for hosts.
3Speed
If data with high access frequency is stored in local storage, then access speed is improved, but load on that single server increases beyond threshold
Solution Approach 1:
The volume is divided into multiple areas that can be distributed across different servers. High-access-frequency data segments are placed on servers with available capacity, distributing the load while maintaining fast local access for each segment.
Solution Approach 2:
Different areas of the volume can be placed on different servers based on their access patterns and the servers' current load conditions. This allows optimal local caching of frequently accessed data while preventing any single server from becoming overloaded.
4Productivity
If volume areas are distributed across multiple nodes, then performance scalability is improved, but system complexity for managing load distribution increases
Solution Approach 1:
The storage system automatically monitors server load conditions and dynamically migrates volume areas between servers based on current capacity and performance requirements. This self-managing approach enables performance scaling without proportional increases in manual management complexity.
Solution Approach 2:
The system dynamically adjusts the distribution of volume areas across servers based on real-time load conditions. This dynamic behavior allows the system to automatically optimize performance and load distribution without requiring complex static configuration or manual intervention.
Data Source
AI summary
A storage system including a processor performs a process for providing a volume including a plurality of areas to one or more hosts. One or more storage devices are connected to the processor and store data of the volume, in which each of a plurality of nodes monitors a load of the volume provided by an own node and loads of areas obtained by dividing an area of the volume into a plurality of the areas, and a first node determining that the load of the one volume being monitored is equal to or more than a threshold value migrates a portion of the area included in the one volume to a volume of a second node different from the first node according to the load of the areas obtained by dividing the area of the one volume into the plurality of areas and a policy of load sharing.