Aggregation Service Hierarchies for Cloud Host Scalability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing scheduler systems in cloud environments face scalability issues when managing multiple host computers, as querying each host for resource information does not efficiently scale with increasing numbers of hosts.
Innovation Solution
A computer-implemented method for creating and managing aggregation service hierarchies, where host computers are configured as leaf, branch, or root nodes based on predefined heuristics and maximum branching factors, allowing dynamic reassignment in response to host additions and failures to maximize fault tolerance and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single process or machine queries each host computer for resource information, then complete resource information can be obtained, but the system does not scale well as the number of hosts increases
Solution Approach 1:
The system segments the monolithic querying process into a hierarchical structure where host computers are organized into trees with root nodes, intermediate nodes, and leaf nodes. Each node manages a subset of hosts and aggregates resource information locally, eliminating the need for a single process to query all hosts directly. This segmentation enables the system to scale to thousands of hosts while maintaining complete resource information through distributed aggregation.
2Ease of operation
If host computers are organized in a flat structure, then simple management is achieved, but efficiency decreases when querying resource information across many hosts
Solution Approach 1:
The system transforms the flat, two-dimensional host management structure into a three-dimensional hierarchical tree structure. Hosts are organized under parent-child relationships with multiple levels of intermediate nodes, adding a vertical dimension to the management architecture. This hierarchical arrangement maintains operational simplicity through automated node selection algorithms while dramatically improving query efficiency by limiting the scope of resource information gathering to relevant subtrees.
3Stability of the object's composition
If the hierarchy structure is fixed, then stable configuration is maintained, but adaptability decreases when hosts are added or fail
Solution Approach 1:
The system implements dynamic hierarchy management where the tree structure automatically reconfigures in response to host additions, removals, or failures. Node selection algorithms continuously evaluate current system state and adjust parent-child relationships, host assignments, and tree topology to maintain optimal balance and fault tolerance. This dynamic behavior allows the hierarchy to adapt to changing conditions while preserving configuration stability through consistent rebalancing rules and automated recovery procedures.
Data Source
AI summary
Techniques disclosed herein provide an approach for managing aggregation service hierarchies. In some embodiments, a hierarchy of an aggregation service is identified. The hierarchy comprises a plurality of nodes, where a respective node is associated with at least one host computer. The aggregation service places resource consumers based on the nodes. A host computer is assigned as a child host of a leaf node based on a clustering heuristic. The clustering heuristic requires the host computer to have access to at least one resource that is accessible to an existing child host of the leaf node. A resource consumer associated with the leaf node is executed on the host computer.