Failover System Using WORM Key-Value Store for Data Restoration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Failover events in primary datacenters often result in data loss and require costly scaling of backup systems, leading to complications such as buffer overflows and bandwidth issues, which disrupt application services like navigation and transportation services.
Innovation Solution
A failover system with real-time replication components that communicate with user devices to restore active data sessions by utilizing a WORM key-value store model and versioning, allowing seamless failover to backup datacenters without extensive backend replication.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If backup systems are scaled to handle failover events, then service availability is improved, but system complexity and cost increase
Solution Approach 1:
The patent uses WORM (Write Once Read Many) key-value store model to create immutable copies of data at the backup datacenter. This allows the backup system to store historical data without requiring complex real-time synchronization mechanisms, reducing system complexity while maintaining service availability during failover events.
Solution Approach 2:
The system performs preliminary data replication to the backup datacenter before failover events occur. By pre-positioning data in the backup location using simple WORM storage, the system avoids the need for complex real-time data synchronization during critical failover moments, thereby improving reliability without proportionally increasing system complexity.
2Loss of information
If real-time data replication is implemented, then data consistency is improved, but bandwidth consumption increases
Solution Approach 1:
The patent extracts only the essential data needed for failover restoration and stores it in a compact WORM key-value format at the backup datacenter. This selective data extraction approach maintains sufficient data consistency for service restoration while minimizing the bandwidth required for data transfer compared to full real-time replication of all data.
Solution Approach 2:
The system uses inexpensive WORM storage objects at the backup datacenter that are optimized for historical data retention rather than real-time updates. This approach accepts that backup data may not be perfectly current but is sufficient for restoration purposes, thereby reducing bandwidth consumption while maintaining adequate data consistency for failover scenarios.
3Loss of information
If extensive backend replication is performed, then data completeness is improved, but restoration time increases
Solution Approach 1:
The system performs preliminary replication of essential data to the backup datacenter using WORM storage before failover events occur. This pre-positioning of data ensures that when failover is needed, the backup datacenter already has the necessary data locally available, eliminating the need for time-consuming data transfer during restoration and reducing overall restoration time while maintaining data completeness.
Solution Approach 2:
The patent creates compact WORM copies of data at the backup datacenter that contain all necessary information for service restoration. These pre-created copies eliminate the need for extensive real-time replication during failover events, thereby maintaining data completeness while significantly reducing restoration time compared to systems that must replicate data during the failover process.
Data Source
AI summary
A failover system can receive active data from user devices running an application specific to a service entity providing an application service. For each user device, the active data can indicate a current status. The failover system can transmit restoration data to the user devices for storage to restore the current status of the user devices in the case of a failover event. When a failover event occurs, the failover system can recover the restoration data from a first user device to restore the current status of the application service for the first user device.