Hierarchical Resource Management for Server Cluster Capacity Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current server cluster virtualization management systems, such as VMware DRS, are limited in managing resource allocation across large data centers with multiple server groups, leading to suboptimal resource management and difficulty in adding capacity to overloaded groups.
Innovation Solution
A hierarchical resource management system that monitors and automatically adjusts the number of host systems across server clusters or groups, using a capacity balancer to move underutilized hosts to overloaded areas, ensuring balanced resource allocation without compromising isolation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If current DRS implementations are used to manage server clusters, then resource management within a single group of N servers can be balanced, but resource management across multiple server groups or large data centers cannot be optimized
Solution Approach 1:
The patent divides the server management system into hierarchical segments: individual server clusters managed by DRS, and multiple such clusters organized into server groups. Each cluster maintains its own resource balancing, while the hierarchical structure enables coordinated management across groups, allowing the system to scale from single-cluster to multi-cluster deployments without losing optimization capability
Solution Approach 2:
The patent extends resource management from a single-dimension (within one cluster) to multi-dimensional management by organizing clusters into groups and enabling cross-cluster resource allocation. This hierarchical dimensionality allows DRS to operate at multiple levels simultaneously, managing resources both within individual clusters and across the broader group structure
2Reliability
If server clusters are limited to N servers per group, then DRS can manage resources within each group, but adding capacity to overloaded groups becomes difficult
Solution Approach 1:
The patent establishes hierarchical resource management structures and cross-cluster resource sharing mechanisms in advance, so that when a cluster becomes overloaded, pre-configured pathways and policies are already in place to rapidly allocate resources from underutilized clusters, enabling quick capacity expansion without structural reconfiguration
Solution Approach 2:
The patent implements dynamic resource allocation where cluster boundaries and resource assignments are not fixed but can change based on demand. Hosts can be dynamically migrated between clusters, and resource pools are adjusted in real-time based on utilization metrics, allowing the system to adapt capacity distribution as workloads change
3Reliability
If multiple server groups are operated with different configurations or customer dedications, then customer isolation is maintained, but optimized resource sharing between groups cannot be achieved
Solution Approach 1:
The patent introduces a hierarchical management layer that acts as an intermediary between isolated customer groups. This intermediate structure enables resource coordination and sharing across groups while maintaining the isolation boundaries, allowing resources to be optimized at the group level without compromising customer-specific security or configuration requirements
Data Source
AI summary
A resource management system for a virtual machine computing environment includes a software component that optimizes capacity between server clusters or groups by monitoring the capacity of server clusters or groups and automatically adding and removing host systems to and from server clusters or groups. The software component may be implemented at a server cluster management level to monitor and execute host system moves between server clusters and/or at a higher level in the resource management hierarchy. At the higher level, the software component is configured to monitor and execute host system moves between sets of server clusters being managed by different server cluster management agents.