Dedicated Network Paths for Cloud Compute Instances
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users of cloud provider networks face challenges in achieving dedicated networking paths for high-performance workloads, as existing technologies often rely on shared network links, leading to network contention and reduced performance.
Innovation Solution
A hardware virtualization service is configured to enable users to request dedicated networking paths between compute instances, with the control plane sending signals to configure routing information and dedicate network links, ensuring exclusive network capacity for the workload.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If shared network links are used to connect compute instances, then network resource utilization and flexibility are improved, but network contention increases and performance deteriorates
Solution Approach 1:
The patent segments the network into shared network links and dedicated network links. Compute instances can be connected via shared links for general traffic or via dedicated links for high-performance workloads. This segmentation allows the system to provide both resource utilization efficiency and performance reliability by choosing the appropriate link type based on workload requirements.
2Reliability
If dedicated network links are allocated to compute instances, then network performance and reliability are improved, but network complexity and resource allocation overhead increase
Solution Approach 1:
The hardware virtualization service automatically manages the allocation and configuration of dedicated network links. When a compute instance requests high-performance networking, the service self-service allocates dedicated links without requiring manual network configuration. This eliminates the complexity burden from users while maintaining performance benefits.
Solution Approach 2:
The hardware virtualization service acts as an intermediary between compute instances and the physical network infrastructure. It abstracts the complexity of dedicated link management by handling routing information updates and link allocation automatically, allowing users to request dedicated networking without dealing with underlying network complexity.
3Loss of energy
If shared network infrastructure is used, then resource efficiency and cost-effectiveness are improved, but network security and isolation between workloads are reduced
Solution Approach 1:
The patent applies different network quality levels to different workloads. Critical high-performance workloads receive dedicated network links with enhanced security and isolation, while non-critical workloads use shared links for resource efficiency. This local quality differentiation allows the system to optimize security and performance for specific workloads without sacrificing overall resource efficiency.
Data Source
AI summary
Techniques are described for enabling users of a cloud provider network to launch compute instances with the option to dedicate networking paths between the compute instances in an otherwise shared networking environment. A user can send, for example, a request to launch two or more compute instances, where the request further indicates a request for a dedicated networking path between the instances. A hardware virtualization service launches the requested compute instances and further sends signals to the computing devices at which the instances are launched indicating the request for a dedicated networking path. A computing device, upon receiving such a signal for a hosted instance, configures routing information at the computing device to dedicate, to the identified instance, an available network link accessible to the computing device and further propagates the signal to additional networking devices as needed to form a dedicated networking path between the associated compute instances.