SVC Cluster Configuration Node Failover via External Subscription Storage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing SAN Volume Controller (SVC) clusters face issues with configuration node failures, leading to loss of client subscription information and event notifications, as the new configuration node does not retain this information, resulting in disrupted service and missed event notifications for subscribed clients.
Innovation Solution
A failover system and method that activates a new configuration node in response to failures, where client subscription information and storage device event logs are stored outside the original configuration node, allowing seamless transition and continued event notifications to clients, with the new node retrieving and sending event information based on subscription details.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If client subscription information and event logs are stored in the original configuration node, then the system structure is simple, but data loss occurs during configuration node failures
Solution Approach 1:
The patent extracts client subscription information and event log storage from the original configuration node and places them in a separate, persistent storage location that is independent of any single node. This extraction ensures that when a configuration node fails, the critical data remains intact and can be accessed by the new configuration node, thereby preventing data loss while maintaining a relatively simple overall system structure.
Solution Approach 2:
The patent introduces an intermediary storage mechanism that acts as a mediator between configuration nodes and client subscriptions. This intermediary storage layer decouples the data from the specific configuration node, allowing seamless failover while preserving data integrity. The intermediary serves as a buffer that maintains the relationship between events and subscriptions without being tied to any single node's lifecycle.
2Productivity
If a new configuration node is activated immediately upon failure, then service continuity is maintained, but client subscription information is lost
Solution Approach 1:
The patent implements preliminary action by pre-storing client subscription information and event logs in a persistent, node-independent location before any failure occurs. This preliminary preparation ensures that when a configuration node fails and a new one is activated, the subscription information is already available, eliminating the need for re-subscription and preventing information loss while maintaining service continuity.
Solution Approach 2:
The patent provides beforehand cushioning by creating a protective storage layer for subscription information that cushions against the impact of node failures. This cushioning mechanism ensures that even if a node fails immediately, the subscription data is protected and remains accessible, allowing the new configuration node to resume service without losing client subscriptions.
3Reliability
If event notifications are sent from the original configuration node, then event tracking is accurate, but notifications are interrupted during failover
Solution Approach 1:
The patent applies universality by making event log storage and subscription management functions independent of any specific configuration node. The persistent storage mechanism serves multiple nodes universally, allowing any configuration node to access and send event notifications to subscribed clients. This multi-functional approach ensures that event tracking remains accurate while notification delivery continues uninterrupted during failover, as the new node can immediately assume the notification function.
Data Source
AI summary
An SVC cluster manages a plurality of storage devices and includes a plurality of SVCs interconnected via a network, each SVC acting as a separate node. A new configuration node is activated in response to configuration node failures. The new configuration node retrieves client subscription information about events occurring in storage devices managed by the SVC cluster from the storage devices. In response to events occurring in the storage device managed by the SVC cluster, the new configuration node obtains storage device event information from a storage device event monitoring unit. The new configuration node sends storage device events to clients who have subscribed to this information according to subscription information obtained. The storage device is not installed in the original configuration node.