Distributed Storage Node Grouping for Power Saving
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In distributed storage systems using the consistent hash method, simultaneously stopping a large number of storage apparatuses leads to data unavailability and performance degradation due to reduced redundancy and increased power consumption during power-saving operations.
Innovation Solution
A distributed storage system that divides storage nodes into groups based on operating states, creating a specified number of data replicas for each group to maintain availability and performance even when multiple nodes are stopped, by traversing a hash ring to determine replication destinations and ensure the required redundancy level.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If a large number of storage apparatuses are simultaneously stopped for power-saving operations, then power consumption is reduced, but data availability and performance deteriorate due to reduced redundancy
Solution Approach 1:
The patent segments storage apparatuses into multiple groups (first storage apparatus group, second storage apparatus group, etc.) based on their operating states. This segmentation allows selective stopping of specific groups while maintaining operational groups that preserve data redundancy and availability, thus resolving the contradiction between power saving and data availability.
Solution Approach 2:
The patent dynamically adjusts the number of data replicas created for each group based on the group's operating state. When storage apparatuses are stopped, the system dynamically modifies replication configurations to ensure that remaining operational groups maintain sufficient redundancy levels, enabling flexible power management without compromising data availability.
2Use of energy by moving object
If a large number of storage apparatuses are simultaneously stopped, then power consumption is reduced, but system performance degrades due to increased data access latency
Solution Approach 1:
By dividing storage apparatuses into distinct groups with different operating states, the system ensures that client requests can be served by operational groups with appropriate data replicas, maintaining performance while allowing non-operational groups to save power.
Solution Approach 2:
The system performs preliminary actions by pre-configuring replication destinations and maintaining minimum replica counts in operational groups before power-saving operations begin, ensuring that performance requirements are met even when many storage apparatuses are stopped.
3Reliability
If data is stored with high redundancy to maintain availability during node failures, then data availability improves, but power consumption increases due to continuous operation of more storage apparatuses
Solution Approach 1:
The patent dynamically adjusts redundancy levels by modifying the number of data replicas based on the operating state of storage groups. When storage apparatuses are stopped, the system reduces redundancy configurations accordingly, allowing lower power consumption while maintaining sufficient availability through the operational groups.
Solution Approach 2:
The system changes the parameter of replica count per group based on operating state. By adjusting this parameter dynamically, the system achieves optimal balance between data availability and power consumption, avoiding fixed high-redundancy configurations that waste energy when not all storage apparatuses need to be operational.
Data Source
AI summary
Storage nodes participating in the distributed storage system are divided into groups beforehand. A data replica creation unit operates so that a replica is created on each of the nodes participating in the distributed storage system on a hash ring. A unit that specifies an operating state of the system determines the number of data replicas for each group. The data replica creation unit traverses the hash ring and creates one or more replicas until it is determined that the number of replicas for each group is reached.