RAID Randomized Distribution for Balanced Degradation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current RAID storage systems face uneven performance degradation due to the relative location of failed storage devices, leading to unpredictable high or low degradation scenarios, which is undesirable for maintaining system reliability and efficiency.

Innovation Solution

A randomized distribution scheme is implemented for data and syndrome segments across storage devices in a RAID stripe, using a randomization table to ensure that any two failed storage devices have a consistent probability of being protected by the same or different type-1 syndrome segments, regardless of their location, thereby maintaining even degradation and improving system resilience.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If data segments and syndrome segments are distributed in an even manner across storage devices, then load balancing is improved, but performance degradation becomes unpredictable when storage devices fail

Engineering Contradiction:
Improveload balancingVSAvoidperformance degradation predictability
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The patent changes the distribution parameter from fixed even distribution to randomized distribution. By using a randomization table with randomly ordered sequences, the system maintains load balancing while ensuring that the relative positions of failed devices becomes irrelevant, making performance degradation predictable regardless of failure location

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent introduces dynamic randomness into the distribution scheme. The randomization table provides different random sequences for different RAID stripes, creating a dynamic distribution pattern that adapts to each stripe while maintaining overall system balance and predictability

Inventive Principle:
Principle #15Dynamics

2Ease of operation

If syndrome segments are distributed in an even manner across storage devices, then load balancing is improved, but relative location of failed devices affects protection consistency

Engineering Contradiction:
Improveload balancingVSAvoidprotection consistency
Core Design Contradiction:
Ease of operationVSStability of the object's composition

Solution Approach 1:

The patent changes the distribution parameter from fixed even distribution to randomized distribution. The randomization table ensures that syndrome segments are still evenly distributed across devices (maintaining load balance) while eliminating the correlation between device location and protection consistency

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent achieves homogeneity in protection consistency across all storage devices. By using randomized distribution, every device has an equal probability of containing any type of segment, ensuring that no device has a privileged or disadvantaged position regarding data protection

Inventive Principle:
Principle #33Homogeneity

Data Source

PatentUS9436394B2RAID random distribution scheme
Publication Date: 2016.09.06 KAMINARIO TECH
  • US9436394B2 patent drawing
  • US9436394B2 patent drawing
  • US9436394B2 patent drawing

AI summary

There is provided a computerized method, computerized system, a non-transitory program storage device and a computer program product for enabling allocation of segments in a RAID stripe to respective storage devices in a corresponding RAID group. A randomized distribution scheme is generated for randomly distributing the data segments and syndrome segments in a RAID stripe to respective storage devices in the RAID group. The randomized distribution scheme is used for distributing the segments in the at least one RAID stripe to the storage devices in the RAID group, such that any two failed storage devices in the RAID group would have the same first probability (P) for being protected by the same type-1 syndrome segment and the same second probability (1-P) for each one of the two failed storage devices being protected by a different type-1 syndrome segment, irrespective of the relative location of the two failed storage devices.