Inode Metadata Cache for Fast Backup Restoration

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional methods for restoring data from inode-based backups to a path-based structure are time-consuming and inefficient, especially when dealing with large numbers of files, as they require repeated traversal of the file tree to obtain metadata associated with specific inode numbers.

Innovation Solution

The method involves generating a sendmap structure to select relevant inodes, creating a metadata cache to store inode metadata, and using this cache to reduce the need for repeated tree traversals, thereby improving restoration speed and efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional methods traverse the file tree repeatedly to obtain metadata for each inode number, then complete metadata retrieval is achieved, but restoration time becomes extremely slow when there are a large number of files

Engineering Contradiction:
Improvemetadata retrieval completenessVSAvoidrestoration time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies preliminary action by creating an inode map that stores all inode numbers and their corresponding metadata paths before the restoration process begins. This pre-computed mapping structure allows the system to directly retrieve metadata without repeatedly traversing the file tree during restoration, thus achieving complete metadata retrieval while significantly reducing restoration time.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If the file tree is traversed repeatedly for each metadata request, then accurate metadata is obtained, but processing efficiency deteriorates with large numbers of files

Engineering Contradiction:
Improvemetadata accuracyVSAvoidrestoration efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent applies copying by creating a simplified inode map that copies and stores the essential mapping relationship between inode numbers and their metadata paths. This copy of the structural information allows rapid lookup without accessing the original file tree repeatedly, maintaining metadata accuracy while dramatically improving restoration efficiency for large numbers of files.

Inventive Principle:
Principle #26Copying

3Quantity of substance

If inode-based backup structure is used for storage efficiency, then storage space is optimized, but restoration complexity increases when converting to path-based structure

Engineering Contradiction:
Improvestorage space utilizationVSAvoidrestoration process complexity
Core Design Contradiction:
Quantity of substanceVSDevice complexity

Solution Approach 1:

The patent introduces an intermediary inode map structure that bridges the inode-based backup storage and the path-based restoration target. This intermediary contains pre-computed mappings that translate inode numbers to their corresponding path information, thereby maintaining storage efficiency while significantly simplifying the restoration process by eliminating the need for complex repeated tree traversals.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentUS11340999B2Fast restoration method from inode based backup to path based structure
Publication Date: 2022.05.24 EMC IP HLDG CO LLC
  • US11340999B2 patent drawing
  • US11340999B2 patent drawing
  • US11340999B2 patent drawing

AI summary

One embodiment is related to a method for restoring an inode based backup to a path based structure, comprising: receiving a path to be restored; determining one or more selected inodes, wherein metadata associated with the one or more selected inodes is to be used in the restoration; generating a sendmap structure that is indicative of the selected inodes; restoring a first file; and determining whether an inode number of the first file, which identifies an inode of the first file, and its associated metadata are available in a metadata cache, and if so: using the inode number of the first file and its associated metadata in the metadata cache in the restoration of the first file.