Active/Active Appliance Cluster Self-Balancing Failover
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing high-availability clusters for active/active appliance clusters rely on external systems like centralized databases and load balancers, which are inefficient for data synchronization and maintenance, and lack scalability, especially for n-node deployments.
Innovation Solution
A method and apparatus that utilize two standby redundant groups within the cluster, including a transaction processing standby group with a self-balancing module and a persistent storing standby group, where the primary database appliance sends 'Hello' messages with priority fields to determine preemptive failover, allowing the standby database appliance to take over duties without external dependencies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If external systems like centralized databases and load balancers are used for high-availability clusters, then data synchronization and maintenance efficiency improve, but device complexity and scalability worsen
Solution Approach 1:
The patent extracts the high-availability functionality from external systems and embeds it within the appliance cluster itself. Each appliance maintains local state information and can independently participate in failover decisions, eliminating the need for external centralized databases and load balancers while maintaining synchronization efficiency.
Solution Approach 2:
The appliance cluster performs self-management for high availability through internal mechanisms. Appliances exchange state information directly with each other and autonomously execute failover procedures without requiring external coordination systems, making the cluster self-sufficient and reducing external dependencies.
2Reliability
If external load balancers and centralized storage are used, then high availability is achieved, but scalability to n-node deployments worsens due to performance impact
Solution Approach 1:
The patent segments the centralized high-availability management into distributed individual appliance decisions. Each appliance independently monitors cluster state and can initiate failover based on local observations, allowing the system to scale to n nodes without the performance bottleneck of a centralized external system.
Solution Approach 2:
The patent transitions from a centralized external dimension of high-availability management to a distributed internal dimension where each appliance operates autonomously. This dimensional shift from external-centralized to internal-distributed architecture enables scalable n-node deployments while maintaining high availability.
3Reliability
If appliances rely on external systems for state management, then transaction persistence is maintained, but ease of operation and maintenance worsen
Solution Approach 1:
The patent merges the state management functionality from external systems into the appliances themselves. Each appliance maintains its own state information locally and exchanges this state directly with other cluster members, combining persistence and synchronization operations into the appliance units for easier maintenance and operation.
Data Source
AI summary
A method executes a preempt by a standby database appliance in a high-availability active/active appliance cluster. The appliance cluster includes a transaction processing standby group and a persistent storing standby group. The transaction processing standby group includes a primary active appliance and a standby appliance. One or more processors receive a Hello message from the primary DB appliance. The processor(s) examine a priority field in the Hello message, in order to determine a priority of the standby database appliance according to the persistent state to thereby determine whether the standby database appliance requests a preempt, where the persistent state includes a state of an application and a database of the primary DB appliance. The processor(s) implement a failover in response to the preempt request to thereby take over a duty of the primary DB appliance.