Hierarchical Network Locality for Cloud Workload Placement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud infrastructure systems face challenges in achieving optimal throughput for high-performance computing applications due to the lack of flow entropy and the arbitrary selection of host machines without considering locality information, leading to inefficient workload execution.
Innovation Solution
Utilizing hierarchical locality information of host machines within cloud infrastructure to guide workload placement, ensuring that workloads are executed on machines based on their physical location and network neighborhood.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If host machines are selected arbitrarily for workload execution, then device complexity is reduced, but throughput deteriorates
Solution Approach 1:
The system pre-computes and stores hierarchical locality information (rack, switch, data center levels) for each host machine before workload assignment. This preliminary action enables fast, informed workload placement decisions without complex real-time calculations, resolving the contradiction between simplified device complexity and improved throughput.
Solution Approach 2:
The patent applies local quality by making workload placement decisions based on the specific locality characteristics of individual host machines. Instead of uniform arbitrary assignment, the system considers the physical location and network topology of each host, assigning workloads to hosts with favorable locality properties to maximize throughput while maintaining manageable system complexity.
2Productivity
If workload placement ignores locality information, then ease of operation is improved, but productivity deteriorates
Solution Approach 1:
Locality information for all host machines is pre-computed and stored in a readily accessible format before workload assignment occurs. This preliminary preparation maintains ease of operation during workload assignment while enabling productivity improvements through informed placement decisions that consider physical location and network topology.
Solution Approach 2:
The system changes the parameter of workload placement from arbitrary selection to locality-aware selection by incorporating hierarchical location parameters (rack, switch, data center). This parameter change improves throughput without significantly complicating operation, as the locality information is pre-processed and integrated into the placement decision mechanism.
3Productivity
If hierarchical locality information is obtained and used for workload placement, then throughput is improved, but device complexity increases
Solution Approach 1:
The patent segments locality information into a hierarchical structure with distinct levels (rack-level, switch-level, data center-level). This segmentation organizes the complex information into manageable segments that can be processed and stored efficiently, enabling throughput improvement through locality-aware placement without overwhelming system complexity.
Solution Approach 2:
The hierarchical locality information is pre-computed and stored for each host machine before workload assignment. This preliminary action transforms complex real-time information gathering into a simple lookup process during workload placement, improving throughput while keeping device complexity manageable through advance preparation.
Data Source
AI summary
Discussed herein are techniques that utilize hierarchical locality information of host machines included in a cluster network for the execution of general workloads. Hierarchical locality information for each host machine of a plurality of host machines is stored. The hierarchical locality information for a host machine identifying, for each locality of a plurality of localities, location information for the locality. Responsive to receiving a request requesting execution of a workload, the hierarchical locality information for the plurality of host machines is obtained and provided (e.g., to a customer) in response to the request.