Virtual Cluster Resource Scheduler Using Distributed Host Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cluster technologies require a dedicated management infrastructure to monitor and manage resource allocation across host computers in a virtualized environment, leading to inefficiencies and increased resource consumption, particularly when handling overloaded conditions or VM migrations.
Innovation Solution
A peer-to-peer distributed architecture using a message bus for inter-host communication, where each host independently determines its load, memory, and IO scores to autonomously manage resource distribution and migrate VMs, eliminating the need for a centralized management system and optimizing resource utilization across the cluster.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a dedicated management infrastructure is used to monitor and manage resource allocation, then resource monitoring capability is improved, but system complexity and resource consumption increase
Solution Approach 1:
Each host computer in the cluster autonomously determines its own load, memory, and IO scores without requiring a centralized management system. The hosts self-monitor their resource utilization and independently make decisions about VM migrations based on their calculated scores and the needs of other hosts in the cluster.
Solution Approach 2:
The patent removes the dedicated management infrastructure (hosts, virtual machines, and databases) from the cluster architecture. By extracting this centralized management layer, the system eliminates the complexity and additional resource consumption associated with policing functions, while maintaining resource monitoring through distributed self-assessment by each host.
2Reliability
If a dedicated management infrastructure is deployed, then resource policing function is improved, but network bandwidth consumption increases
Solution Approach 1:
Hosts independently monitor and police their own resource utilization by calculating their own load, memory, and IO scores. This eliminates the need for centralized polling and reporting mechanisms that would consume network bandwidth, as each host autonomously determines its resource state and makes migration decisions based on cluster needs.
Solution Approach 2:
By removing the dedicated management infrastructure that performs policing functions, the patent eliminates the network communication overhead associated with centralized resource monitoring. The policing function is distributed to individual hosts, reducing network bandwidth consumption while maintaining resource management capability.
3Ease of operation
If centralized management infrastructure is used, then resource allocation control is improved, but system overhead increases
Solution Approach 1:
Each host autonomously calculates its resource scores and determines its own allocation needs based on VM requirements and cluster conditions. This distributed self-determination eliminates the overhead of centralized control mechanisms while maintaining effective resource allocation through autonomous host decision-making.
Solution Approach 2:
The resource allocation control function is segmented and distributed to individual hosts rather than centralized. Each host independently manages its own resource allocation decisions based on its calculated scores and cluster conditions, reducing the overhead associated with centralized control while maintaining allocation effectiveness.
Data Source
AI summary
In some embodiments, a method for cluster resource scheduling, includes determining at least one load score; determining a memory score; determining an IO score; and monitoring a message bus for candidate messages when each of the at least one load score, memory score, and IO score is less than a pre-determined health threshold. In some embodiments, a host computer system for hosting a plurality of virtual machines (VMs), includes: a memory; a network adapter for communicating with the cluster by way of a message bus; a processor in electronic communication with the memory and the network adapter, wherein the processor is programmed to: determine at least one load score; determine a memory score; determine an IO score; and monitor the message bus for candidate messages when each of the at least one load score, memory score, and IO score is less than a pre-determined health threshold.