DNS Latency Estimation via TCP Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
DNS lookups contribute significantly to perceived latency during Internet browsing, and existing methods struggle to accurately measure and reduce latency across different protocols, particularly due to the inability of UDP to calculate round trip times.
Innovation Solution
A system and method that access DNS request data via TCP, select a representative sample, estimate latency, and modify network configurations based on estimated latency, using techniques like random sampling and bootstrapping to account for confidence intervals and traffic volume, while also considering potential anycasting and peer group rankings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If DNS lookup operations are performed to translate domain names into IP addresses, then domain name resolution is achieved, but significant latency is introduced during Internet browsing
Solution Approach 1:
The patent performs DNS lookups in advance before they are actually needed for web browsing. By proactively resolving domain names to IP addresses and caching the results, the system eliminates the need for time-critical DNS queries during user browsing activities, thus reducing perceived latency while maintaining resolution accuracy
Solution Approach 2:
The patent creates copies of DNS resolution results and stores them in cache memory. Instead of performing repeated DNS lookup operations for the same domain names, the system uses cached copies of IP address mappings, dramatically reducing the time required for subsequent domain name resolutions without sacrificing accuracy
2Measurement precision
If TCP protocol is used to calculate round trip time for latency measurement, then accurate latency data can be obtained, but UDP protocol cannot provide round trip time calculations
Solution Approach 1:
The patent introduces TCP as an intermediary measurement protocol to estimate UDP latency characteristics. By sending TCP packets through the same network path and measuring round trip times, the system obtains accurate latency data that serves as a proxy for UDP performance, enabling latency measurement without requiring actual UDP timing capabilities
Solution Approach 2:
The patent changes the measurement parameter from direct UDP round trip time (which is unavailable) to TCP round trip time (which is available). This parameter substitution allows the system to derive meaningful latency information using a different but correlated metric, achieving both measurement precision and protocol versatility
3Reliability
If a large volume of DNS request data is collected from multiple subnets, then comprehensive latency analysis is possible, but processing complexity increases
Solution Approach 1:
The patent extracts and separates different components of the DNS request data, organizing them by subnet, domain name, and timing characteristics. By decomposing the large dataset into manageable segments with specific attributes, the system can process and analyze each component independently, reducing overall processing complexity while maintaining comprehensive analysis capabilities
Solution Approach 2:
The patent divides the comprehensive DNS request data into discrete segments organized by subnet and time period. This segmentation allows the system to process smaller, manageable portions of data in parallel or sequentially, reducing memory requirements and computational complexity while preserving the ability to perform reliable aggregate latency estimation across all subnets
Data Source
AI summary
Systems, methods, and computer-readable mediums are provided that access a set of data related to a plurality of domain name system (DNS) requests for a plurality of subnets in a network. A subset of the set of data that is a representative sample of the set of data is selected. Latency of the subset of the data is estimated and latency is estimated for the totality of the data. A portion of the network is modified based on the estimated latency of the totality of the data.