Virtual Switch QoS Adaptation for VM Migration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional virtual switches provide limited Quality of Service (QoS) options, mainly restricting bandwidth control for ingress interfaces, failing to offer adaptable and differentiated services for virtual machines (VMs) across various network environments.
Innovation Solution
A virtual switch with a multiplane architectural framework, including an adaptation plane, switching plane, control plane, application plane, and management plane, supports dynamic allocation of queues and multiple simultaneous controllers, enabling selective QoS tools like classification, congestion management, congestion avoidance, bandwidth limiting, priority manipulation, and traffic shaping, which can migrate with VMs across servers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional virtual switches provide QoS control, then bandwidth limiting for ingress interfaces is achieved, but the QoS options are limited and cannot provide differentiated services for multiple VMs
Solution Approach 1:
The virtual switch is divided into multiple independent queues, each capable of being assigned to different VMs or traffic types. This segmentation allows differentiated QoS services for multiple VMs simultaneously while maintaining manageable complexity through modular queue management
Solution Approach 2:
The system dynamically allocates and reassigns queues to different VMs based on real-time requirements. VMs can be migrated between queues, and queue assignments can be changed without disrupting network connectivity, providing adaptable QoS that responds to changing workload demands
2Productivity
If VMs are migrated between physical servers, then resource utilization is optimized, but QoS policies and statistics are lost during migration
Solution Approach 1:
QoS policies, queue assignments, and traffic statistics are copied and replicated to the destination server during VM migration. This ensures that the migrating VM retains its QoS characteristics and policy settings on the new physical hardware, maintaining service continuity and quality
Solution Approach 2:
The QoS management system operates universally across multiple physical servers, maintaining consistent policy enforcement regardless of the underlying hardware. The same QoS policies applied on one server can be seamlessly transferred and enforced on another server, providing hardware-independent QoS management
3Adaptability or versatility
If multiple QoS tools are implemented for differentiated services, then service quality for various VMs is improved, but the system complexity increases
Solution Approach 1:
Differentiated QoS services are implemented through segmentation into separate queues, each handling specific traffic types or VM requirements. This modular approach enables multiple QoS tools (classification, congestion management, bandwidth limiting, priority manipulation, traffic shaping) to operate independently and be selectively applied without overwhelming complexity
Solution Approach 2:
Each queue is configured with local quality parameters specific to its intended purpose, allowing tailored QoS treatment for different VMs or traffic types. This localized configuration approach enables differentiated services while keeping individual queue management simple and manageable
Data Source
AI summary
In one embodiment a method includes selectively controlling a virtual network switch to control a quality of service (QoS) for a flow associated with a virtual machine (VM). Controlling the QoS may include controlling the virtual switch to provide classification, congestion management, congestion avoidance, bandwidth limiting, traffic shaping, and/or priority manipulation. The method may also include migrating a policy associated with a virtual interface between the VM and the virtual switch from a first server to a second server when the VM is to be migrated from the first server to the second server. The migration may also include moving statistics associated with the policy. The method may include selectively adapting the policy or an existing policy on the receiving server when a discrepancy is detected between the policies. The method may include reserving resources to provide guaranteed minimum bandwidth, even without control of an ingress and egress queue.