DNS Reflection GTM Latency Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for Global Traffic Management (GTM) are inaccurate due to their reliance on passive and active techniques that do not account for dynamically changing network conditions, and they require compatible Application Delivery Controllers (ADCs) across data centers, limiting efficient traffic distribution among geographically distributed data centers.
Innovation Solution
A method that uses technology-independent DNS reflection methods in combination with a shared database to estimate Round Trip Times (RTTs) between Local Domain Name Servers (LDNSs) and data centers, allowing for optimal traffic routing and message handling decisions, regardless of ADC type or vendor, and periodically updates RTT metrics to account for changing network conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional active probing techniques are used to estimate network latency, then latency estimates can be obtained, but network congestion increases and many data centers do not respond to probes
Solution Approach 1:
The patent introduces DNS servers as intermediary elements to measure network latency. Instead of direct probing between data centers, the system uses DNS query-reflection mechanisms where DNS servers act as mediators to indirectly measure latency by timing DNS query responses. This intermediary approach obtains latency data without generating harmful network congestion.
Solution Approach 2:
The patent replaces conventional active probing mechanisms (ICMP pings, TCP SYN packets) with DNS-based measurement mechanisms. By substituting the mechanical probing system with a DNS query system, the patent achieves latency measurement while leveraging the existing DNS infrastructure that is already present in the network, thereby avoiding additional congestion.
2Productivity
If conventional GTM methods use vendor-specific ADCs, then traffic distribution can be managed, but compatibility and interoperability between different data centers are limited
Solution Approach 1:
The patent implements a universal DNS-based measurement mechanism that works across multi-vendor environments. By using DNS protocols (which are standardized and vendor-neutral) instead of vendor-specific ADC probing mechanisms, the system achieves universal compatibility while maintaining efficient traffic distribution. The DNS infrastructure serves multiple purposes: domain resolution and latency measurement.
Solution Approach 2:
The patent changes the measurement parameter approach from vendor-specific ADC metrics to standardized DNS query timing metrics. By altering the measurement methodology to use DNS protocol parameters (query send time, response receive time) instead of vendor-specific ADC parameters, the system achieves interoperability across different vendors while maintaining measurement accuracy.
3Ease of operation
If passive geographic location techniques are used for GTM, then simple data center selection is achieved, but dynamic network conditions are not accounted for
Solution Approach 1:
The patent performs preliminary DNS-based latency measurements and stores the results in a database before actual traffic distribution decisions are made. This preliminary action captures dynamic network conditions in advance, allowing the system to use both the pre-measured latency data and geographic location information for more accurate data center selection without adding complexity to real-time operations.
Solution Approach 2:
The patent implements a feedback mechanism where DNS-based latency measurements are continuously performed and stored in a database. This feedback loop provides up-to-date network condition information that can be used to adjust traffic distribution decisions, combining the simplicity of geographic-based selection with the accuracy of real-time latency data.
Data Source
AI summary
An example method facilitates Global Traffic Management (GTM) and associated selective distribution of computing loads and/or network traffic among different geographically distributed data centers of a particular domain by allocating the servicing of request messages from Local Domain Name Servers (LDNSs) to data centers that are estimated to provide the quickest response times to the LDNSs. Estimates of path latencies or Round Trip Times (RTTs) between the LDNSs and the different distributed data centers are estimated using DNS reflection methodologies and are maintained in a database that may be accessed and collectively managed via the data centers, e.g., via use of a web service that fronts the database. Locations of the LDNSs approximate locations of client devices seeking access to the domain.