Host-Based RAID Storage Partitioning for Compute Blade Density

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional compute blade chassis designs with local hard disk mirroring for fault tolerance impact density, form factor, airflow, thermal cooling, and cost, requiring a minimum of two hard disk drives per blade, leading to inefficiencies in storage and fault recovery.

Innovation Solution

A host-based RAID solution is implemented in a shared storage environment where compute blades are coupled to concentrators, with shared disk drives configured in a RAID array, allowing each compute blade to access dedicated partitions across multiple drives, enabling a RAID 0+1 configuration for fault tolerance and reduced drive requirements.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If each compute blade has local hard disk drives for fault tolerance, then reliability is improved, but device complexity and cost increase

Engineering Contradiction:

Improvefault toleranceVSAvoiddrive configuration

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent extracts the storage function from individual compute blades and consolidates it into a shared storage resource. Instead of each blade having local drives, storage is externalized to a common pool that multiple blades access, simplifying the overall system architecture while maintaining fault tolerance through redundancy.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The shared storage resource serves multiple compute blades simultaneously, providing a universal storage solution that replaces individual blade-specific storage systems. This multi-functional approach allows a single storage infrastructure to support multiple computing units, reducing total drive count and system complexity.

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Reliability

If each compute blade has local hard disk drives, then fault tolerance is improved, but compute blade density decreases

Engineering Contradiction:

Improvefault toleranceVSAvoidcompute blade density

Core Design Contradiction:

ReliabilityVSVolume of moving object

Solution Approach 1:

Storage drives are extracted from the compute blade form factor and relocated to a separate shared storage resource. This separation allows compute blades to be designed without integrated drives, increasing blade density in the chassis while fault tolerance is maintained through the redundant shared storage architecture.

Inventive Principle:

Principle #2Taking out (Extraction)

3Reliability

If each compute blade has local hard disk drives, then fault tolerance is improved, but thermal cooling performance worsens

Engineering Contradiction:

Improvefault toleranceVSAvoidthermal cooling

Core Design Contradiction:

ReliabilityVSTemperature

Solution Approach 1:

Heat-generating storage drives are extracted from compute blades and relocated to a dedicated shared storage resource with its own thermal management. This separation improves compute blade thermal performance by removing heat sources, while the shared storage system can be cooled independently.

Inventive Principle:

Principle #2Taking out (Extraction)

4Reliability

If each compute blade has local hard disk drives, then fault tolerance is improved, but air flow efficiency worsens

Engineering Contradiction:

Improvefault toleranceVSAvoidair flow obstruction

Core Design Contradiction:

ReliabilityVSObject-generated harmful factors

Solution Approach 1:

Physical storage drives are extracted from compute blades and consolidated into a shared storage resource. This removal eliminates obstructions to air flow across compute blade surfaces, improving cooling efficiency, while fault tolerance is maintained through the redundant shared storage configuration.

Inventive Principle:

Principle #2Taking out (Extraction)

5Reliability

If each compute blade has local hard disk drives, then fault tolerance is improved, but system cost increases

Engineering Contradiction:

Improvefault toleranceVSAvoidnumber of drives

Core Design Contradiction:

ReliabilityVSQuantity of substance

Solution Approach 1:

Individual blade storage drives are merged into a consolidated shared storage resource pool. Instead of each blade having dedicated drives, multiple blades share a common set of drives with redundancy, reducing the total number of drives required while maintaining fault tolerance through the RAID 0+1 configuration.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The shared storage resource provides universal storage capacity to multiple compute blades, replacing numerous individual blade-specific drives with a single multi-functional storage system. This consolidation reduces drive quantity and system cost while maintaining reliability.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS7653782B2Method for host bus adapter-based storage partitioning and mapping across shared physical drives

Publication Date: 2010.01.26 DELL PROD LP

AI summary

A system for a host-based RAID solution in a shared storage environment is provided in which the compute blades of a system are coupled to one or multiple concentrators. The concentrators serve as a switch or expander to couple each of the compute blades to a shared storage resource. Within the shared storage resource, a set of drives is configured in a RAID array. The shared disk drives are partitioned so that each partition is dedicated to one of the compute blades of the system. Multiple sets of drives may be used so that the collective set of drives can be configured as part of a RAID volume that includes mirroring between at least two of the drives of the RAID volume, such as RAID 1 or RAID 0+1, in which each set of drives is a mirror of the other set of drives and the content associated with each of the compute blades is striped across multiple drives in each of the two sets of drives.