Data Deduplication Rehydration Locality Optimization

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

As the number of backup images increases in data deduplication systems, the data segments referenced by latter images become scattered across multiple data containers, worsening locality and leading to decreased rehydration performance, which is the process of reading and restoring files.

Innovation Solution

The method involves identifying data containers with data segments referenced by a backup image, determining if the number of segments in a container is below a certain threshold, and rebasing those segments to another container to decrease the total number of data containers, thereby improving locality and rehydration performance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If the number of backup images increases in data deduplication systems, then the storage capacity and backup capability are improved, but the data segments referenced by backup images become scattered across multiple data containers, worsening locality and leading to decreased rehydration performance

Engineering Contradiction:
Improvenumber of backup imagesVSAvoidrehydration performance
Core Design Contradiction:
Quantity of substanceVSProductivity

Solution Approach 1:

The patent merges data segments from multiple scattered data containers into a consolidated structure. By combining references to data segments across different containers and reorganizing them into a unified view, the system improves locality without limiting the number of backup images. This merging approach allows backup images to access their referenced data segments more efficiently, resolving the contradiction between storing many backup images and maintaining fast rehydration performance.

Inventive Principle:
Principle #5Merging (Combining)

2Adaptability or versatility

If data segments are distributed across multiple data containers to accommodate growing backup images, then the storage system can handle more backup images, but the locality of data segments deteriorates and rehydration time increases

Engineering Contradiction:
Improvecapacity to handle backup imagesVSAvoidrehydration time
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent implements preliminary action by pre-organizing and pre-positioning data segment references before rehydration operations are needed. The system maintains an organized mapping structure that anticipates rehydration needs, allowing data segments to be quickly located and assembled even when distributed across multiple containers. This preparatory organization minimizes rehydration time while preserving the ability to handle growing numbers of backup images.

Inventive Principle:
Principle #10Preliminary action

3Quantity of substance

If the number of data containers increases to store data segments from more backup images, then the system can support more backup images, but the amount of time needed to read and restore files increases

Engineering Contradiction:
Improvenumber of data containersVSAvoidfile restoration speed
Core Design Contradiction:
Quantity of substanceVSProductivity

Solution Approach 1:

The patent implements feedback mechanisms that monitor data segment distribution and rehydration performance across data containers. The system uses this feedback to dynamically optimize the organization of data segments, identifying patterns in how segments are accessed and redistributed to improve locality. This continuous feedback loop allows the system to maintain fast file restoration speeds even as the number of data containers increases to support more backup images.

Inventive Principle:
Principle #23Feedback

Data Source

PatentUS10002050B1Systems and methods for improving rehydration performance in data deduplication systems
Publication Date: 2018.06.19 COHESITY INC
  • US10002050B1 patent drawing
  • US10002050B1 patent drawing
  • US10002050B1 patent drawing

AI summary

The disclosed computer-implemented method for improving rehydration performance in data deduplication systems may include (1) identifying a set of data containers that include data segments referenced by a backup image stored in a data deduplication system, (2) determining the total number of data segments referenced by the backup image that are included in a data container within the set of data containers, (3) determining that the total number of data segments that are referenced by the backup image and included in the data container is below a certain threshold, and then (4) rebasing the data segments that are referenced by the backup image and included in the data container to at least one other data container to decrease the total number of data containers that include the data segments referenced by the backup image. Various other methods, systems, and computer-readable media are also disclosed.