Decoupled Resource Allocation in Distributed Computing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In distributed computing systems, conflicts between multiple schedulers attempting to use the same computing resource lead to inefficient scheduling, negatively impacting workload throughput, resource utilization, and runtime performance, especially as the system scales and workload volume increases.
Innovation Solution
Implementing a method where resource offer managers transmit resource offers to schedulers, allowing them to request and allocate resources based on state, including availability and sharing policies, with synchronized resource acquisition cycles to manage resource allocation and scheduling independently across multiple schedulers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple schedulers are used to manage different workloads and resources, then workload management capability is improved, but scheduling conflicts increase and efficiency deteriorates
Solution Approach 1:
The patent introduces a resource manager as an intermediary component that mediates between multiple schedulers and computing resources. The resource manager receives resource requests from schedulers, resolves conflicts centrally, and allocates resources accordingly, preventing scheduling conflicts before they occur while maintaining the benefits of multiple specialized schedulers
2Quantity of substance
If the number of workloads increases to handle more tasks, then system capacity is improved, but scheduling complexity and time consumption increase
Solution Approach 1:
The patent implements preemptive resource allocation where the resource manager proactively assigns resources to schedulers based on predicted workload patterns and resource availability. By performing resource allocation in advance and maintaining a pool of pre-available resources, the system reduces scheduling decision time when workload volume increases
3Ease of operation
If schedulers independently allocate resources, then scheduling autonomy is improved, but resource allocation conflicts increase
Solution Approach 1:
The patent establishes a feedback mechanism where schedulers report resource requests and allocation status to the resource manager, which responds with conflict resolution decisions and allocation confirmations. This closed-loop feedback system maintains scheduler autonomy while ensuring centralized coordination to prevent allocation conflicts
Data Source
AI summary
Systems and methods are disclosed herein for allocating computing resources for execution of workloads in a distributed computing system with resource acquisition cycles and workload scheduling cycles. One or more resource offer managers transmit resource offers to each of one or more of schedulers. Each scheduler transmits a resource request and/or resource release request in response to each resource offer manager. Each resource offer manager computes a resource allocation for each scheduler after resolving conflicts and assigns the resources to each scheduler according to the resource allocation decisions. The resource acquisition cycle and the workload scheduling cycle are decoupled.