Metadata-Based File System Cloning for Faster Application Migration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing cloud bursting process for application migration in storage systems is delayed due to the time-consuming data migration of large data volumes, affecting user experience.
Innovation Solution
A file system cloning method that creates a cloned file system in the destination storage system using only metadata, without transferring user data, and supports incremental updates and data prefetching to reduce delays.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all data in the local storage system is migrated to the public cloud before application deployment, then the application can run on the cloud storage system, but the data migration takes a long time and causes processing delay
Solution Approach 1:
The patent creates a cloned file system in advance in the destination storage system using only metadata from the source file system, without waiting for complete data migration. This preliminary action enables the application to be deployed and started immediately on the cloud, while data is transferred incrementally in the background as needed.
Solution Approach 2:
The patent creates a cloned file system by copying only the metadata structure (directory entries, file information, allocation tables) from the source to the destination, rather than copying all actual data. This lightweight copying approach allows the cloned file system to be created rapidly and enables the application to run on the cloud while data is transferred on-demand.
2Loss of time
If a cloned file system is created using only metadata without data transfer, then the application migration delay is reduced, but the cloned file system initially lacks actual data content
Solution Approach 1:
The destination storage system performs preliminary actions by creating the cloned file system structure with metadata in advance, enabling immediate application deployment. The actual data transfer is deferred to occur incrementally in the background as the application accesses files, ensuring data availability without delaying migration.
Solution Approach 2:
The patent implements continuous data transfer in the background after the cloned file system is created. As the application runs and accesses files on the cloud, data is continuously transferred from the local storage system to the cloud storage system, ensuring that data content becomes available progressively without interrupting application operation.
3Productivity
If metadata is transferred to create the cloned file system, then the cloning process is fast and application migration delay is reduced, but the amount of data transferred is minimal which may not be sufficient for complete file system replication
Solution Approach 1:
The patent segments the file system replication process into two distinct phases: (1) rapid metadata cloning that creates the file system structure immediately, and (2) incremental data transfer that occurs in the background. This segmentation allows the cloning speed to be high while the total data transfer volume is managed over an extended period without blocking application deployment.
Solution Approach 2:
The metadata cloning is performed as a preliminary action that establishes the complete file system structure quickly. This preliminary metadata transfer enables fast cloning speed and immediate application deployment, while the actual data transfer is deferred to occur incrementally as needed, balancing speed with complete replication.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A file system cloning method and apparatus is provided. In the method, a destination storage system first receives first information from a source storage system in which a file system runs. The first information is used to indicate a data layout of the file system. Then, the destination storage system creates a cloned file system of the file system based on the first information. After the destination storage system creates the cloned file system corresponding to the file system in the source storage system, an application running on the file system in the source storage system can also run on the cloned file system. Only a data layout of a source file system is required for creating the cloned file system. In other words, data of a file included in the file system does not need to be transmitted, and an amount of data in the data layout is less than an amount of data included in the file. In this way, a delay in an application migration process can be reduced.