Dual-Rail Fabric Placement for VM Migration in HPC Clouds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing high performance computing environments face challenges in efficiently managing network traffic and live migration of virtual machines due to complex addressing and routing schemes, particularly in InfiniBand networks, which hinder scalability and flexibility.
Innovation Solution
Implementing a heterogeneous and asymmetric dual-rail fabric configuration that utilizes prepopulated LIDs in a vSwitch architecture, allowing for dynamic reconfiguration and efficient resource utilization, while supporting live migration of virtual machines.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional network and storage architectures are used in cloud computing, then administrative simplicity is maintained, but performance bottlenecks occur
Solution Approach 1:
The network fabric is segmented into multiple independent rails (e.g., rail 0 and rail 1), each capable of carrying traffic independently. This segmentation allows the system to achieve higher aggregate bandwidth by utilizing multiple separate network paths, thereby improving productivity without requiring a complete architectural overhaul.
Solution Approach 2:
The dual-rail fabric architecture provides multi-functionality by supporting both traditional network traffic and high-performance storage traffic simultaneously across different rails. The system can dynamically allocate rails for different purposes, making the network infrastructure universally applicable to various workloads while maintaining administrative simplicity.
2Adaptability or versatility
If complex addressing and routing schemes are implemented in InfiniBand networks, then high performance is achieved, but scalability and flexibility are hindered
Solution Approach 1:
A fabric manager acts as an intermediary between the complex InfiniBand addressing/routing mechanisms and the virtual machine migration process. The fabric manager handles the complexity of address translation and route determination, presenting a simplified interface to the migration system while maintaining the high-performance capabilities of the underlying InfiniBand network.
Solution Approach 2:
The complexity of addressing and routing is extracted from the virtual machine migration process and isolated in dedicated fabric management components. This extraction allows the migration system to focus on mobility management while the fabric manager handles the complex network addressing and routing, thereby improving scalability without sacrificing performance.
3Reliability
If virtual machine live migration is performed in traditional networks, then downtime occurs, but network resources are underutilized
Solution Approach 1:
Network resources are pre-configured and reserved for potential virtual machine migrations before they are needed. The dual-rail fabric maintains standby capacity and pre-established routes that can be immediately activated during migration events, eliminating the need for resource provisioning during the migration process and thereby reducing downtime while ensuring continuous service availability.
Data Source
AI summary
Systems and methods for supporting heterogeneous and asymmetric dual rail fabric configurations in a high performance computing environment. A method can provide, comprising at one or more computers each including one or more microprocessors, a plurality hosts, each of the plurality of hosts comprising at least one dual port adapter, a private fabric, the private fabric comprising two or more switches, and a public fabric, the public fabric comprising a cloud fabric. A workload can be provisioned at a host of the plurality of hosts. A placement policy can be assigned to the provisioned workload. Then, network traffic between peer nodes of the provisioned workload can be assigned to one or more of the private fabric and the public fabric in accordance with the placement policy.