Distributed Hypervisor Load Balancer for Data Center Packet Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Centralized health-check mechanisms in data centers face resource constraints when polling numerous servers, leading to inefficiencies and potential network disruption detection failures as they cannot effectively monitor all servers simultaneously.
Innovation Solution
A decentralized health-monitoring mechanism where health monitoring modules on each host device collect and share health statistics with local load balancing modules, enabling efficient load balancing of network packets between virtual computing instances within the data center.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a centralized health-check mechanism is used to monitor server health, then load balancing decisions can be made, but the mechanism faces resource constraints and cannot effectively monitor all servers simultaneously
Solution Approach 1:
The centralized health-check mechanism is segmented into distributed health-monitoring modules deployed on individual host devices. Each module independently monitors health metrics for virtual computing instances on its local host, eliminating the single-point bottleneck and enabling parallel monitoring across all servers in the data center.
Solution Approach 2:
Host devices and virtual computing instances perform self-monitoring of their own health metrics without requiring external polling. Each health-monitoring module collects local health information autonomously and shares it with the data center, enabling the system to monitor itself distributedly rather than requiring centralized resource expenditure.
2Productivity
If health information is collected and shared in a decentralized manner, then monitoring efficiency improves, but the system complexity increases
Solution Approach 1:
While health monitoring is decentralized, the patent merges health information collection and sharing functionality into unified health-monitoring modules that operate on each host. These modules standardize the collection, processing, and sharing of health metrics, reducing the complexity that would otherwise arise from entirely independent distributed agents.
Solution Approach 2:
Health-monitoring modules act as intermediaries between local virtual computing instances and the broader data center load-balancing system. They aggregate and standardize health information from multiple sources and distribute it appropriately, simplifying the complexity of direct peer-to-peer health information exchange across the distributed system.
3Ease of operation
If traditional load balancers are used for East-West traffic, then load balancing can be performed, but they create single points of failure and bottlenecks
Solution Approach 1:
The traditional centralized load balancer is segmented into distributed load-balancing modules deployed on individual host devices. Each module independently performs load balancing for East-West traffic on its local host, eliminating single points of failure and distributing the load-balancing function across multiple nodes in the data center.
Solution Approach 2:
Instead of traffic flowing through centralized load balancers that act as intermediaries, the patent inverts the architecture so that load-balancing intelligence resides at the edge on host devices. This reverses the traditional client-server load balancing model and enables direct peer-to-peer load balancing between virtual computing instances on different hosts.
4Ease of operation
If centralized load balancers are used, then load balancing decisions can be made, but they create bottlenecks that reduce packet forwarding efficiency
Solution Approach 1:
The centralized load-balancing control function is segmented into distributed load-balancing modules on each host. This segmentation enables parallel processing of load-balancing decisions across multiple hosts simultaneously, eliminating the sequential bottleneck of centralized control and increasing overall packet forwarding throughput.
Solution Approach 2:
The patent transitions from a single-dimensional centralized control architecture to a multi-dimensional distributed architecture where load-balancing decisions occur across multiple spatial dimensions (different hosts, different modules). This dimensional expansion enables concurrent load-balancing operations that significantly increase packet forwarding speed.
Data Source
AI summary
The disclosure provides an approach for load balancing packets within a data center. The approach leverages dynamically collected and up-to-date health information on each virtual computing instance located within the data center. In one embodiment, health monitoring modules, located within hypervisors of each host computer, collect health statistics on local virtual computing instances. Each health monitoring module shares its locally collected health statistics with every other health monitoring module. Each health monitoring module provides the shared health statistics, on all virtual computing instances within the data center, to a local load balancing module located within the hypervisor of each host computer. Each load balancing module uses health statistics of all virtual computing instances to load balance packets within the data center. Further, the disclosure describes an affinity-based load balancing approach in which a local load balancing module may give preference to local virtual computing instances when making load balancing decisions.