Cross-Protocol Live Data Migration via SPDK Transport
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data storage network migration processes are often disruptive, causing application downtime and inefficiencies, especially when transitioning from spinning-disk storage to NVM storage arrays, as they require manual data copying and are not non-disruptive, predictable, or scalable.
Innovation Solution
A non-disruptive cross-protocol live data migration method that uses the Storage Performance Development Kit (SPDK) to establish a transport session, create a namespace based on SSD characteristics, and utilize multipath settings to route I/Os seamlessly between storage arrays, ensuring no application downtime and supporting heterogeneous environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual data copying is used for storage network migration, then data can be transferred from old storage network to new storage network, but application downtime occurs and migration is disruptive
Solution Approach 1:
The system performs preliminary actions by establishing a transport session and creating namespace mappings before actual data migration begins. The multipath I/O configuration is prepared in advance, allowing the system to switch paths seamlessly during migration without application downtime.
Solution Approach 2:
The patent introduces an intermediary mechanism - the SPDK transport session and namespace layer - that mediates between the application and the underlying storage arrays. This intermediary allows data to be migrated from old to new storage arrays while the application continues to access data through the namespace, eliminating disruptive downtime.
2Adaptability or versatility
If traditional data migration methods are used, then data transfer between storage arrays can be achieved, but the process is not scalable and requires proprietary software
Solution Approach 1:
The system employs universal interfaces and protocols through SPDK that enable the migration solution to work across heterogeneous storage environments without requiring proprietary software. The transport session mechanism provides multi-functionality by supporting various storage protocols and array types, enhancing adaptability while reducing complexity.
3Productivity
If live data migration is implemented, then application downtime is eliminated, but I/O routing between multiple storage arrays becomes complex
Solution Approach 1:
The system implements self-service through automated I/O routing where the multipath layer automatically manages path selection and failover between old and new storage arrays. The namespace configuration enables automatic I/O redirection without requiring complex manual routing rules, allowing live migration to proceed efficiently with reduced configuration complexity.
Data Source
AI summary
A non-disruptive cross-protocol live data migration process, such as from a legacy SAN storage network to a modern NVMe-oF storage network, includes maintaining a bitmap to track the transferred blocks at migration time, and utilizing the bitmap for intelligent I/O routing decisions that take into account characteristics of SSD-type NVM storage, such as endurance characteristics associated with solid state type memory. Furthermore, protocol command translations and multipath layer capabilities such as I/O queuing are advantaged at migration time to hide the underlying complexities related to a cross-protocol mass data migration, to enable a non-disruptive data migration process intended to avoid application downtime.