Online Upgrade Sequence for Distributed Storage Nodes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In distributed storage systems, the online upgrade process is prolonged due to the need for each storage node to wait for service data to be migrated back from a backup node before the next node can be upgraded, leading to increased overall upgrade time.
Innovation Solution
A method where a monitoring node sets an upgrade sequence for storage nodes and determines backup nodes based on this sequence, allowing service data to be stored on nodes that have already completed their upgrades, enabling subsequent nodes to start their upgrades without waiting for data migration to complete.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a storage node waits for service data to be migrated back from a backup node before the next node can be upgraded, then data integrity is ensured, but the overall upgrade time increases
Solution Approach 1:
The patent applies preliminary action by pre-establishing backup node relationships and pre-migrating service data before the upgrade sequence begins. The monitoring node determines backup nodes in advance based on the upgrade sequence, and service data is migrated to backup nodes before each storage node is upgraded, eliminating the need to wait for data migration during the upgrade process.
Solution Approach 2:
The patent implements continuity of useful action by enabling parallel execution of upgrade operations and data migration processes. Multiple storage nodes can be upgraded simultaneously while data migration occurs in the background, ensuring that the upgrade process continues without interruption rather than waiting for each migration to complete.
2Productivity
If service data is temporarily written onto a backup node during online upgrade, then the current node can be upgraded, but the backup node must wait to migrate data back before the next upgrade
Solution Approach 1:
The monitoring node determines backup nodes and establishes data migration paths before the upgrade sequence begins. Service data is pre-migrated to backup nodes before each storage node enters upgrade mode, so that when a node needs to be upgraded, the data is already in place and no waiting is required.
Solution Approach 2:
The patent introduces dynamic scheduling of upgrade operations and data migration processes. The monitoring node dynamically adjusts the upgrade sequence and data migration timing based on system state, allowing flexible parallel execution of multiple operations rather than rigid sequential processing.
3Stability of the object's composition
If multiple storage nodes are upgraded sequentially with data migration between upgrades, then system stability is maintained, but the total upgrade duration increases
Solution Approach 1:
The monitoring node pre-determines the complete upgrade sequence and backup node assignments before execution. Data migration is initiated in advance for each node in the sequence, ensuring that when a node is upgraded, its data is already migrated to the appropriate backup node, eliminating wait time while maintaining sequential stability.
Solution Approach 2:
The patent segments the upgrade process into independent parallel tasks: data migration tasks and upgrade execution tasks. These segmented tasks can run concurrently with different timing, allowing the system to maintain sequential upgrade stability while reducing total duration through parallel background operations.
Data Source
AI summary
In a method, identifiers are obtained of N to-be-upgraded nodes. An upgrade sequence of the N to-be-upgraded nodes is set according to the identifiers of the N to-be-upgraded nodes. A backup node of each to-be-upgraded node is determined according to the upgrade sequence of the N to-be-upgraded nodes, to obtain node backup information. A backup node of the Xth to-be-upgraded node is at least one node in a set that includes a to-be-upgraded node prior to the Xth to-be-upgraded node and a to-be-upgraded node following the (X+[Z+1])th to-be-upgraded node. Z is a ratio of a time required by the Xth to-be-upgraded node for migrating back service data to an online upgrade time of the Xth to-be-upgraded node. The node backup information is sent to the N to-be-upgraded nodes.