Host Admission Control via Future Reservation Slots
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern large data centers face revenue loss due to high utilization rates leading to disproportionate blocking of virtual machines (VMs) with largest resource requirements, and existing solutions like priority settings and static reservations fail to adequately manage capacity and revenue optimization in dynamic environments.
Innovation Solution
Implementing a Future Reservations system that categorizes resource requests based on characteristics like resource requirements and revenue, setting thresholds to block less preferred deployments when utilization is high, and reserve capacity for future preferred deployments, thereby maximizing resource utilization and revenue.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If high utilization rates are maintained to maximize resource usage, then resource efficiency is improved, but revenue is lost due to blocking of high-resource VM deployments
Solution Approach 1:
The system performs preliminary actions by setting aside future reservation slots in advance before they are needed. The admission control component monitors current utilization and proactively reserves capacity for future high-value VM deployments, preventing revenue loss while maintaining current resource efficiency.
Solution Approach 2:
The system changes the parameter of resource allocation by introducing dynamic blocking thresholds and future reservation slots. When utilization exceeds thresholds, the system transitions from accepting all requests to selectively blocking non-critical requests, thereby preserving revenue while maintaining efficiency.
2Reliability
If static reservations are implemented to reserve capacity for future deployments, then future revenue is protected, but current resource utilization decreases
Solution Approach 1:
Instead of reserving excessive capacity that would reduce current utilization, the system applies partial action by reserving only the minimum necessary future slots. The blocking thresholds are set to allow maximum current utilization while still guaranteeing future deployment capacity.
Solution Approach 2:
The system makes the reservation system dynamic rather than static. The admission control component continuously monitors utilization and adjusts blocking decisions in real-time, allowing the system to adapt between maximizing current utilization and protecting future revenue based on actual conditions.
3Loss of energy
If priority settings are used to favor high-revenue VMs, then revenue optimization is improved, but system complexity increases
Solution Approach 1:
The system optimizes revenue by changing the parameter of request acceptance through dynamic blocking thresholds. Instead of complex priority queuing, the system uses simple threshold-based admission control that automatically favors high-revenue VMs when thresholds are exceeded, maintaining low complexity while achieving revenue optimization.
Data Source
AI summary
Systems and methods for admission control to a physical host system are provided herein. One aspect provides for receiving at least one resource request at an admission control component of a distributed computing system, the at least one resource request comprised of at least one system type; processing the at least one resource request utilizing at least one physical host accessible to the distributed computing system; specifying a number of resource request slots to be reserved for at least one system type based on at least one future reservation threshold accessible to the admission control component; and blocking resource requests from entering the system through the admission control component based on a number of available resource request slots and the at least one future reservation threshold. Other embodiments and aspects are also described herein.