Sparse Replica File System for Cluster High Availability

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current systems face challenges in maintaining high-availability of file system instances across computing nodes, particularly when one node fails, leading to data unavailability and potential data loss, especially in larger installations with increased storage operations, and are costly due to full replication requirements.

Innovation Solution

Implementing a sparse replica file system instance on a third computing node during failure of a second node, allowing data updates to be directed to the first replica and the sparse replica, and rebuilding the second replica upon recovery, thereby reducing storage and networking overhead.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If full replication is implemented across computing nodes, then high availability is improved, but storage costs and networking overhead increase

Engineering Contradiction:

Improvehigh availabilityVSAvoidstorage costs

Core Design Contradiction:

ReliabilityVSQuantity of substance

Solution Approach 1:

The patent segments the replication approach by implementing a two-tier replication strategy: full replication between primary and secondary nodes, and sparse replication between primary and tertiary nodes. This segmentation allows the system to maintain high availability through multiple replicas while reducing overall storage costs by applying less resource-intensive sparse replication only where needed for failover scenarios.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies local quality by implementing different replication qualities in different system locations: full replication is applied between primary and secondary nodes where complete data redundancy is critical, while sparse replication is applied to tertiary nodes where partial data redundancy suffices for specific failover scenarios. This localized approach optimizes the balance between availability and storage costs.

Inventive Principle:

Principle #3Local quality

2Reliability

If full replication is implemented across computing nodes, then high availability is improved, but networking overhead increases

Engineering Contradiction:

Improvehigh availabilityVSAvoidnetworking overhead

Core Design Contradiction:

ReliabilityVSLoss of energy

Solution Approach 1:

The patent segments the networking approach by implementing different replication communication patterns: full replication networking between primary and secondary nodes, and sparse replication networking between primary and tertiary nodes. This segmentation reduces overall networking overhead by limiting full-duplex communication to only the critical primary-secondary pair while using more efficient sparse synchronization for tertiary nodes.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies local quality by implementing different networking qualities in different system locations: full networking capability is applied between primary and secondary nodes where complete data synchronization is critical, while sparse networking is applied to tertiary nodes where partial data synchronization suffices. This localized approach optimizes the balance between availability and networking resource consumption.

Inventive Principle:

Principle #3Local quality

3Productivity

If computing nodes are added to larger installations, then storage operations increase, but data loss risk increases

Engineering Contradiction:

Improvestorage operationsVSAvoiddata loss risk

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent segments the data protection approach by implementing a three-node architecture with distinct roles: primary node for data operations, secondary node for full replication backup, and tertiary node for sparse replication backup. This segmentation ensures that even in larger installations with increased storage operations, data loss risk is mitigated through multiple replicated copies distributed across segmented nodes with defined failover paths.

Inventive Principle:

Principle #1Segmentation

4Reliability

If sparse replica is created during node failure, then high availability is maintained, but CPU usage increases

Engineering Contradiction:

Improvehigh availabilityVSAvoidCPU usage

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent applies preliminary action by pre-establishing the tertiary computing node with sparse replica capabilities before failure occurs. The sparse replica structure is prepared in advance with the necessary data blocks and metadata, so that upon primary node failure, the tertiary node can immediately assume service responsibilities with minimal additional CPU overhead for data reconstruction or formatting operations.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS11409715B2Maintaining high-availability of a file system instance in a cluster of computing nodes

Publication Date: 2022.08.09 HEWLETT PACKARD ENTERPRISE DEV LP

AI summary

A method for maintaining high-availability of file system instances is described. The method includes maintaining replica file system instances such as a first replica file system instance on a first computing node and a second replica file system instance on a second computing node. Further, a third computing node is instructed to create a sparse replica file system instance on the third computing node in response to detection of a failure condition associated with the second computing node. Moreover, a data update request is directed to the first replica file system instance and the sparse replica file system.