VM Placement Scheduler for Cloud Resource Overbooking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud computing platforms face inefficiencies in resource utilization due to the idle standby redundancy in active-standby cloud applications, leading to wastage of resources and potential failures in VM transitions due to resource contention.
Innovation Solution
A redundancy-aware VM scheduler optimizes the placement of active and standby VMs, ensuring resource entitlements and availability by selectively overbooking standby VMs and minimizing concurrent activations on the same host through careful placement and selection strategies within a hierarchical cloud infrastructure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If standby VMs are kept idle to ensure fault tolerance and disaster recovery, then system reliability is improved, but resource utilization deteriorates with 50% or more of system capacity being wasted
Solution Approach 1:
The patent merges multiple standby VMs onto the same physical host to consolidate idle resources. By co-locating standbys from different applications on shared hosts, the system maintains fault tolerance while improving resource utilization through shared infrastructure.
Solution Approach 2:
The patent introduces a new dimension of resource sharing by allowing standby VMs to share physical hosts dynamically. This transforms the traditional one-to-one mapping between standby and physical resource into a many-to-many relationship, enabling efficient resource pooling across multiple applications.
2Productivity
If standby VMs are overbooked to reduce reserved resources, then resource utilization is improved, but the risk of resource contention during VM activation increases
Solution Approach 1:
The patent implements feedback mechanisms that monitor host resource availability and activation patterns. The system uses this feedback to dynamically adjust overbooking levels and make informed decisions about standby activation, preventing resource contention while maintaining high utilization.
Solution Approach 2:
The patent performs preliminary placement optimization by strategically positioning standby VMs on hosts with sufficient headroom and diverse failure modes. This advance planning ensures that when activation is needed, resources are likely to be available, reducing contention risk before it occurs.
3Productivity
If multiple standby VMs are placed on the same host to improve resource utilization, then resource efficiency is improved, but the chance of simultaneous activation failure increases due to common host failures
Solution Approach 1:
The patent applies local quality by placing standby VMs from different applications on the same host with differentiated resource allocations and isolation mechanisms. This allows efficient co-location while protecting against cascading failures through local resource segmentation and independent activation paths.
Data Source
AI summary
Virtual redundancy for active-standby cloud applications is disclosed herein. A virtual machine (“VM”) placement scheduling system is disclosed herein. The system can compute, for each standby VM of a plurality of available standby VMs, a minimum required placement overlap delta to meet an entitlement assurance rate (“EAR”) threshold. The system can compute a minimum number of available VM slots for activating each standby VM to meet the EAR threshold. For each standby VM of a given application, the system can filter out any server of a plurality of servers that does not meet criteria. If a given server meets the criteria, the system can add the given server to a candidate list; sort, in descending order, the candidate list by the minimum required placement overlap delta and the number of available virtual machine slots; and select, from the candidate list of servers, a candidate server from atop the candidate list.