Zero Downtime Maintenance Bridge Schema for Application Updates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional system update procedures require users to logout and re-login, causing downtime and potential data loss due to the shutdown of application servers, which can be lengthy and disrupt business operations.
Innovation Solution
A zero downtime maintenance (ZDM) procedure is implemented, where a bridge database schema is generated as a copy of the original schema, allowing work processes to seamlessly transition to the new schema without stopping entire application instances, enabling continuous user access and minimizing downtime.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If application servers are stopped and restarted with target DB schema in rolling re-connect, then the system can be updated to a new version, but users who are logged into the application server will need to re-login causing downtime
Solution Approach 1:
The patent segments the monolithic application server into multiple independent application server instances. Each instance can be updated independently without affecting others, allowing gradual migration from old to new DB schema while maintaining user sessions on other instances. This segmentation enables the system to update without forcing all users to re-login simultaneously.
Solution Approach 2:
The patent introduces an intermediary mechanism (session management layer and database connection pooling) that decouples users from specific application server instances. Users are connected to the system through session objects that can be migrated between instances, allowing the underlying infrastructure to change without disrupting user continuity. The session abstraction layer acts as a mediator between users and the evolving backend infrastructure.
2Reliability
If application servers are stopped to ensure data integrity for non-terminated sessions, then data loss is prevented, but the shutdown duration increases to match the longest-running user transaction
Solution Approach 1:
The patent performs preliminary actions by creating a new application server instance with the target DB schema before shutting down the old instance. Data is pre-synchronized and session states are prepared for migration in advance. This allows the system to maintain user sessions continuously while the backend infrastructure is updated, preventing both data loss and lengthy shutdowns.
Solution Approach 2:
The patent ensures continuity of useful action by maintaining user sessions active throughout the update process. Instead of stopping all operations, the system allows users to continue working on the new instance while data synchronization and session migration occur in the background. This continuous operation eliminates the need to wait for longest-running transactions to complete before updating.
3Productivity
If work processes are distributed to multiple application server instances, then system capacity is increased, but the restart time for complete roll-over increases to the sum of time for the longest-running work process per instance
Solution Approach 1:
The patent segments the update process into instance-level independent operations. Each application server instance can be updated, restarted, and migrated independently without waiting for other instances to complete their work processes. This segmentation transforms the sequential roll-over time (sum of all instances) into parallel operations, maintaining high system capacity while reducing total update time.
Solution Approach 2:
The patent applies partial action by updating only one instance at a time rather than coordinating all instances simultaneously. The system accepts that not all instances need to be updated at the exact same moment, allowing incremental progress. This approach reduces the coordination overhead and eliminates the need to wait for the sum of all work process durations, achieving faster roll-over while maintaining distributed architecture benefits.
Data Source
AI summary
The disclosure generally describes methods, software, and systems, including a method for updating an application. At least one application instance of an application is managed. Each application instance is associated with a plurality of executing work processes connected with a first database schema. A bridge database schema is generated that is related to the first database schema. The bridge database schema represents a copy of the first database schema and is generated in response to initiation of an update to the application. In response to determining that the generation of the bridge database schema is complete, for each of the plurality of work processes, a commit work action performed by the particular work process is determined. In response to determining performance of the commit work action, the particular work process is connected to the bridge database schema.