GSLB Controller Round Trip Time Measurement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional GSLB systems face challenges in measuring application response delay due to firewalls and geographical limitations, which can lead to suboptimal server selection and capacity issues, resulting in inefficient load balancing and potential service bottlenecks.
Innovation Solution
A system comprising multiple GSLB controllers that measure round trip times (RTTs) between client DNS servers and application servers, allowing for the calculation of cumulative response times and selection of the server with the lowest delay, even across different geographical locations, thereby optimizing server selection and load balancing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a firewall is placed between the local DNS server and the SLB, then security is improved, but the measurement of server response time becomes impossible
Solution Approach 1:
The patent introduces a firewall penetration tool as an intermediary component that can traverse the firewall to measure RTT between the local DNS server and SLB. This tool acts as a mediator that overcomes the blocking effect of the firewall, allowing response time measurement to continue while maintaining security.
Solution Approach 2:
The patent replaces the traditional direct measurement mechanism with a multi-step indirect measurement approach. Instead of directly measuring RTT through the firewall (which is blocked), the system uses DNS-based indirect measurement to obtain response time information, substituting the mechanical direct measurement process with a network protocol-based indirect measurement.
2Device complexity
If only a single GSLB controller and servers at the same geographical location are used, then system complexity is reduced, but the capacity to handle demand is insufficient
Solution Approach 1:
The patent segments the GSLB system into multiple distributed controllers, each managing servers in different geographical locations. This segmentation allows the system to scale capacity by adding more controllers and servers across different regions, while each individual controller maintains relatively simple architecture.
Solution Approach 2:
The patent transitions from a single-location GSLB system to a multi-location distributed system, adding the geographical dimension to the architecture. This dimensional expansion enables the system to handle increased demand by distributing load across multiple locations while maintaining manageable complexity through modular design.
3Device complexity
If servers co-located with the receiving GSLB controller are used, then device complexity is reduced, but the optimal server selection for client service is not achieved
Solution Approach 1:
The patent makes each GSLB controller universal by enabling it to measure RTT to servers at different geographical locations, not just co-located servers. This multi-functionality allows any controller to evaluate all available servers globally, breaking the limitation that controllers can only manage locally co-located resources.
Solution Approach 2:
The patent implements a feedback mechanism where GSLB controllers continuously measure and exchange RTT information with each other and with local DNS servers. This feedback loop enables controllers to update their knowledge of server performance globally, allowing optimal server selection based on actual measured response times rather than assumptions about geographical proximity.
Data Source
AI summary
A method and system for measuring application response delay is described. The method may commence with receiving a Domain Name System (DNS) request from a client DNS server. The method may include measuring round trip time between the client DNS server and a first Global Server Load Balancing (GSLB) controller, between the first GSLB controller and a server load balancer (SLB) collocated with the first GSLB controller, and between the SLB and an application server. The method may further include receiving measurements of round trip time between the client DNS server and a second GSLB controller, between the second GSLB controller and an SLB collocated with the second GSLB controller, and between the second GSLB controller and a further application server. A cumulative response time associated with the application servers may be calculated based on the measurements to select an application server having a lowest cumulative response time.