Active-Standby Data Center Failover via BGP Conditional Advertisement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In active-standby data center topologies, there is a risk of both data centers becoming simultaneously active, and the standby data center may advertise its IP address unnecessarily, leading to ambiguity in routing and potential service disruptions during failures.
Innovation Solution
Implementing BGP conditional advertisement and Route Health Injection (RHI) to ensure the standby data center only advertises its IP address when the active data center is down, using metrics and community settings to control route advertisement and prevent simultaneous activation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If the standby data center advertises its IP address proactively, then failover speed is improved, but routing ambiguity and simultaneous activation risk increase
Solution Approach 1:
The standby data center performs preliminary health checks on the active data center before activating. It monitors the active site's route advertisements and only begins advertising its own routes after confirming the active site is unavailable, preventing premature activation and routing confusion
Solution Approach 2:
The system implements feedback mechanisms where the standby data center continuously monitors BGP route advertisements from the active data center. When route advertisements stop or health indicators show failure, this feedback triggers the standby to activate and begin advertising its routes, ensuring failover occurs only when necessary
2Reliability
If BGP conditional advertisement is implemented, then simultaneous activation is prevented, but system complexity increases
Solution Approach 1:
The patent introduces Route Health Injection (RHI) as an intermediary mechanism that simplifies BGP conditional advertisement. Instead of complex BGP policy configurations, RHI injects or withdraws routes based on simple health indicators, making the conditional advertisement logic more manageable and less complex
Solution Approach 2:
The data center system performs self-monitoring of its own health status and automatically adjusts route advertisements accordingly. Each data center autonomously determines its own activation state based on monitored health indicators, eliminating the need for external control mechanisms and reducing overall system complexity
Data Source
AI summary
The present invention provides an active/standby data center that avoids the delay associated with a cached DNS entry to switch from the active data center to the standby data center. When the active data center becomes unavailable, the standby data center advertises the same address as the primary data center so the change over occurs quickly. When the IP address of the primary data center is no longer visible to the standby data center, the standby data center begins to advertise.