RAID Hot Extent Migration for Rebuild Performance

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

RAID rebuild processes are hindered by resource competition between data reconstruction and ongoing read/write operations, leading to prolonged rebuild times and increased risk of permanent data loss due to impaired I/O performance.

Innovation Solution

The method involves identifying hot extents in a failed RAID, migrating them to an underused normal RAID, and rebuilding the failed RAID, thereby reallocating resources to focus on reconstruction while minimizing the impact on I/O performance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of time

If more resources are dedicated to rebuilding the RAID, then the rebuild time is reduced, but the I/O performance suffers

Engineering Contradiction:
Improverebuild timeVSAvoidI/O performance
Core Design Contradiction:
Loss of timeVSProductivity

Solution Approach 1:

The patent extracts hot extents (highly accessed data portions) from the RAID undergoing rebuild and relocates them to a different RAID. This separation removes the I/O workload associated with hot extents from the rebuild process, allowing the rebuild to proceed faster without being constrained by ongoing I/O operations to those extents.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent introduces a intermediary RAID as a temporary storage location for hot extents during the rebuild process. This intermediary RAID acts as a mediator that accepts I/O operations for hot extents, thereby protecting the rebuild process from performance degradation while maintaining system I/O capabilities.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Productivity

If more resources are dedicated to servicing I/O requests, then the I/O performance is maintained, but the rebuild time is extended

Engineering Contradiction:
ImproveI/O performanceVSAvoidrebuild time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent extracts hot extents from the RAID undergoing rebuild and relocates them to a different RAID. This separation removes the I/O workload associated with hot extents from the rebuild process, allowing the rebuild to proceed faster without being constrained by ongoing I/O operations to those extents.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent introduces a intermediary RAID as a temporary storage location for hot extents during the rebuild process. This intermediary RAID acts as a mediator that accepts I/O operations for hot extents, thereby protecting the rebuild process from performance degradation while maintaining system I/O capabilities.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Reliability

If the RAID is rebuilt expeditiously by allocating maximum resources, then the risk of permanent data loss is reduced, but the read/write response time is negatively impacted

Engineering Contradiction:
Improvedata loss riskVSAvoidread/write response time
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent extracts hot extents (highly accessed data portions) from the RAID undergoing rebuild and relocates them to a different RAID. This separation removes the I/O workload associated with hot extents from the rebuild process, allowing the rebuild to proceed faster without being constrained by ongoing I/O operations to those extents.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent introduces a intermediary RAID as a temporary storage location for hot extents during the rebuild process. This intermediary RAID acts as a mediator that accepts I/O operations for hot extents, thereby protecting the rebuild process from performance degradation while maintaining system I/O capabilities.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentUS8719619B2Performance enhancement technique for raids under rebuild
Publication Date: 2014.05.06 INTERNATIONAL BUSINESS MACHINE CORPORATION
  • US8719619B2 patent drawing
  • US8719619B2 patent drawing
  • US8719619B2 patent drawing

AI summary

A method for improving the performance of a RAID under rebuild is disclosed. In one embodiment, such a method includes identifying a RAID requiring rebuild, such as by identifying a RAID having one or more failed storage-drive components. The method then automatically performs the following in response to identifying the RAID: the method identifies hot extents (i.e., extents most heavily accessed) in the RAID; the method migrates the hot extents from the identified failed RAID to a normal RAID not requiring rebuild, such as to an underused RAID; and the method rebuilds the failed RAID. The migration of the hot extents will ideally occur while the RAID is being rebuilt but may also be performed prior to the rebuild process. A corresponding apparatus and computer program product are also disclosed.