Storage System Failover via Segmented Virtual Machines

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing techniques for failover in virtual file servers when a physical server fails often require significant resources from the destination server, potentially increasing the load and risking resource imbalances.

Innovation Solution

A storage system with multiple physical servers, where protocol virtual machines and file system virtual machines are created, allowing for resource allocation and failover control, enabling efficient transfer of file system VMs and protocol VMs between servers to manage resource distribution and load balancing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If failover is performed in units of virtual file servers, then the virtual file server can be restored after a physical server failure, but the failover destination physical server requires a considerable amount of resources and the load on the destination server increases

Engineering Contradiction:

Improvefailover capabilityVSAvoidresource requirements

Core Design Contradiction:

ReliabilityVSQuantity of substance

Solution Approach 1:

The virtual file server is segmented into two independent virtual machines: a protocol virtual machine (handling client protocol processing) and a file system virtual machine (handling file management). This segmentation allows selective failover of only the file system VM to a destination server, while the protocol VM remains on the original or a different server, thereby reducing the resource burden on the failover destination server compared to migrating the entire virtual file server.

Inventive Principle:

Principle #1Segmentation

2Reliability

If failover is performed in units of virtual file servers, then service continuity is maintained, but the load on the failover destination physical server increases significantly

Engineering Contradiction:

Improveservice continuityVSAvoidserver load

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

By separating the protocol VM and file system VM, only the file system VM needs to be migrated during failover, reducing the processing load and resource consumption on the destination server while maintaining service continuity.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The protocol VM and file system VM can be combined on different physical servers to distribute the load. The protocol VM handles client communications while the file system VM handles file operations, allowing parallel processing and better load distribution across the infrastructure.

Inventive Principle:

Principle #5Merging (Combining)

Data Source

PatentUS11994965B2Storage system, failover control method, and recording medium

Publication Date: 2024.05.28 HITACHI VANTARA LTD

AI summary

Failover is performed appropriately when a failure occurs in a physical server. In a plurality of physical servers of a storage system, one or more protocol VMs and one or more file system VMs are created. The protocol VMs perform processing related to a protocol for a file storage with a client via a front-end network. The file system VMs perform processing related to management of files in the file storage. A first physical server causes, when a failure occurs in a second physical server, a physical server other than the second physical server to operate a file system VM to be operated in place of the file system VM of the second physical server, and controls the protocol VM of the physical server other than the second physical server to perform processing to be performed by the protocol VM of the second physical server.