Global Site Selector Mesh Probing via Regional Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current global server load balancer systems face increased load and connectivity-dependent probe answer status issues, limiting mesh size expansion without drastic load increases, as each global server load balancer must probe each Virtual IP address.
Innovation Solution
Implementing a primary and standby global site selector configuration where the primary shares Virtual IP address status with other site selectors, allowing incremental status pulls and redundant status derivation from both primary and standby, enabling efficient keepalive management and mesh expansion without excessive load.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If each global server load balancer probes each Virtual IP address in a mesh configuration, then connectivity status can be monitored, but the load on the server load balancer increases drastically
Solution Approach 1:
The system divides the mesh into hierarchical segments with regional servers that aggregate probe results. Instead of every global server probing every Virtual IP directly, regional servers segment the probing task and consolidate status information, reducing the overall probe count and server load while maintaining comprehensive connectivity monitoring.
Solution Approach 2:
Regional servers act as intermediary components between global servers and Virtual IPs. These intermediaries perform the actual probing and return aggregated status information to global servers, eliminating the need for direct many-to-many probing relationships and significantly reducing the load on server load balancers.
2Adaptability or versatility
If the global server load balancer mesh size is increased, then coverage and redundancy improve, but the load on the server load balancer increases drastically
Solution Approach 1:
The mesh is segmented into regional units with local servers handling probes for their respective regions. This segmentation allows the overall mesh size to expand without proportionally increasing the load on central server load balancers, as each regional segment independently manages its own probing workload.
Solution Approach 2:
The system transitions from a flat, direct mesh topology to a hierarchical structure with multiple dimensions (global servers, regional servers, and Virtual IPs). This dimensional change allows larger mesh sizes to be supported by distributing the probing load across multiple hierarchical levels rather than concentrating it on server load balancers.
3Ease of operation
If probe answer status depends on network connectivity between global server load balancer and Virtual IP address, then simple probing is possible, but false 'Offline' answers occur when connection is down
Solution Approach 1:
Regional servers serve as intermediaries that perform the actual connectivity probes to Virtual IPs and return results to global servers. This separation ensures that global servers don't directly depend on network connectivity to Virtual IPs for status determination, eliminating false 'Offline' answers caused by transient connectivity issues between global servers and Virtual IPs.
Solution Approach 2:
The system performs preliminary connectivity assessments through regional servers that have established network paths to Virtual IPs. By pre-establishing these connectivity relationships and using regional servers as permanent proxies for probing, the system avoids the reliability issues of ad-hoc connectivity checks from any global server to any Virtual IP.
Data Source
AI summary
Systems and methods are disclosed for designating a network device as a primary device for probing a virtual IP address and designating a second network device as a standby device for probing the virtual IP address. Results of probe requests may be shared with a plurality of network devices within a mesh.