Stretched Cluster Migration for Zero-Downtime VMs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current active/active storage systems are limited to metropolitan-scale distances, making it impossible to implement seamless data migration between data centers located over 100 kilometers apart without reverting to active/passive architectures, which results in downtime during migration.
Innovation Solution
Implementing a distributed virtual volume system with a stretched cluster mechanism and relay stations to enable synchronous data migration across long distances, using protocols like VPLEX Metro and VMware vMotion, allowing for zero-downtime migration of active virtual machines by encapsulating their state and transferring it over high-speed networks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If active/active storage systems are used for data migration, then data availability and zero-downtime migration are achieved, but the geographic distance is limited to metropolitan-scale (100 kilometers or less)
Solution Approach 1:
The system segments the long-distance data migration path into multiple metropolitan-scale segments, each handled by a relay station. Virtual machine state is transferred through intermediate relay stations that maintain active/active storage relationships, breaking down the impossible long-distance direct connection into manageable short-distance segments that can each support synchronous replication
Solution Approach 2:
Relay stations serve as intermediary systems between source and destination data centers. These relay stations maintain active storage relationships with both previous and next stations in the chain, acting as mediators that enable state transfer across geographic distances that would otherwise exceed the capabilities of active/active storage protocols
2Length of stationary object
If active/passive storage systems are used for long-distance data migration, then geographic distance limitations are overcome, but data downtime occurs during migration
Solution Approach 1:
The system performs preliminary actions by establishing active storage relationships and pre-configuring relay stations before migration begins. Virtual machine state is prepared and staged at relay stations in advance, allowing the actual migration to occur without downtime when the virtual machine is actively moved through the relay chain
3Manufacturing precision
If synchronous replication is used for active/active storage, then data consistency is maintained, but the maximum transmission distance is limited to metropolitan scales
Solution Approach 1:
The long-distance transmission path is segmented into multiple short-distance synchronous replication links between adjacent relay stations. Each segment maintains data consistency through synchronous replication, while the chain of segments collectively spans much larger geographic distances than a single synchronous link could achieve
Data Source
AI summary
A method of providing migration of active virtual machines by performing data migrations between data centers using distributed volume and stretched cluster mechanisms to migrate the data synchronously within the distance and time latency limits defined by the distributed volume protocol, then performing data migrations within data centers using local live migration, and for long distance data migrations on a scale or distance that may exceed synchronous limits of the distributed volume protocol, combining appropriate inter- and intra-site data migrations so that data migrations can be performed exclusively using synchronous transmission.