Compute Instance Migration to Isolated Virtual Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Migrating applications from one logical partition to another in provider networks without affecting application availability is challenging, especially when transitioning from older partitions with limited features to newer partitions with advanced capabilities like isolated virtual networks.
Innovation Solution
Implementing an automated migration service using a migration manager that leverages existing virtualized load balancing and auto-scaling services to orchestrate the transfer of compute instances across partition boundaries, employing techniques like zero-down-time migration, live migration, and reboot migration to minimize downtime.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If applications are migrated from older logical partitions to newer logical partitions with advanced features, then access to advanced features (e.g., isolated virtual networks) is improved, but application availability deteriorates due to migration downtime
Solution Approach 1:
The patent creates equivalent compute instances in the destination logical partition before terminating instances in the source partition. This preliminary action ensures that replacement instances are ready and configured in advance, enabling seamless migration without application downtime. The migration manager orchestrates this by launching equivalent instances, copying storage devices, and establishing network connections before switching traffic to the new instances.
2Ease of operation
If manual migration processes are used to transfer compute instances between logical partitions, then migration control is improved, but migration time and complexity worsen
Solution Approach 1:
The patent implements an automated migration manager that performs the entire migration process without manual intervention. The migration manager automatically discovers compute instances, creates equivalent instances in the destination partition, copies storage devices, configures network settings, and terminates source instances. This self-service automation eliminates manual migration steps while maintaining full control over the migration process, significantly reducing migration time and complexity.
3Adaptability or versatility
If compute instances are terminated in the source logical partition to complete migration, then feature access is improved, but application continuity deteriorates during the transition
Solution Approach 1:
The patent uses a load balancer as an intermediary to maintain application continuity during migration. The load balancer distributes traffic to both source and destination compute instances, allowing seamless switching between them. When instances are terminated in the source partition, the load balancer automatically redirects traffic to the equivalent instances in the destination partition, ensuring no interruption in application service and maintaining continuous availability.
Data Source
AI summary
A migration manager (MM) identifies a source group of compute instances (at a first partition of a provider network) to be migrated to an isolated virtual network (IVN) implemented in a second partition. The MM triggers the establishment within the IVN of an equivalence group of instances corresponding to the source group instances. Connectivity between the equivalence group and a first load balancer which was being used for the source group in the first partition is maintained by modifying one or more configuration settings. After a replacement load balancer is established within the IVN, the first load balancer is decommissioned.