CDN DNS Resolution via Tracker-Based Load Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing content delivery network (CDN) solutions struggle to quickly identify the end point or content server that can best serve an end user in a geographically distributed network, relying heavily on hierarchical DNS servers and hardware load balancers, which limits the efficiency of content delivery.
Innovation Solution
A method for DNS resolution in a CDN service where the ISP DNS resolver sends requests to an authoritative sub-zone DNS server, which forwards them to a tracker that participates in DNS resolution, allowing end points to perform consistent hashing on the requested URL and location to determine the best datacenter or end point to serve the content, and sends a redirect message to the user, enabling direct connection to the least-loaded end point.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If hierarchical DNS servers and hardware load balancers are used to identify content servers, then system reliability is maintained, but content delivery speed and efficiency deteriorate
Solution Approach 1:
The patent extracts the load balancing function from traditional hardware load balancers and hierarchical DNS servers, implementing it directly within end points through consistent hashing algorithms. This removes the bottleneck of centralized resolution and enables direct, fast connection to appropriate content servers without relying on intermediate hardware or complex DNS hierarchies.
Solution Approach 2:
End points are empowered to perform their own load balancing and content server identification using consistent hashing on URL and location data. Each end point independently determines the best content server without requiring external DNS resolution or hardware load balancer intervention, achieving self-service content delivery.
2Measurement precision
If hierarchical DNS servers are used for resolution, then comprehensive content server identification is achieved, but resolution time and system complexity increase
Solution Approach 1:
The system pre-computes and stores mapping relationships between URL hashes, geographic locations, and content servers in distributed data structures at end points. When a content request arrives, the end point directly queries these pre-computed structures using consistent hashing, eliminating the need for time-consuming hierarchical DNS resolution while maintaining accurate content server identification.
3Ease of operation
If hardware load balancers are deployed, then traffic distribution is achieved, but device complexity and cost increase
Solution Approach 1:
The patent replaces mechanical hardware load balancers with software-based consistent hashing algorithms running on distributed end points. This substitution eliminates the need for specialized hardware infrastructure while maintaining effective traffic distribution capabilities through algorithmic content server selection based on URL and location characteristics.
Data Source
AI summary
Comprises identifying an end point or content server that can best serve an end user that sent a DNS request to an ISP DNS resolver, given a geographically distributed network of end points. In particular, the method further comprises using the end points themselves and a tracker to identify and notify to the end user the IP addresses of the least-loaded and closest end points that can best serve the content request.