NPIV Client Crash Prevention via Hypervisor Resource Suspension

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Client logical partitions in a computer system using NPIV (N_Port ID Virtualization) face execution crashes when the NPIV server becomes unavailable, as they attempt to access physical storage through an unavailable virtual input/output server partition, leading to I/O-related crashes.

Innovation Solution

A hypervisor firmware layer dynamically suspends the allocation of virtual processor resources to the client logical partition upon detection of events that would cause a crash, such as the unavailability of the virtual I/O server partition, thereby preventing access attempts to physical storage and ensuring the client partition remains operational during server downtime.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the client logical partition continues to access physical storage through the virtual I/O server partition, then the client partition can maintain normal I/O operations, but the client partition will crash when the server partition becomes unavailable

Engineering Contradiction:

Improveclient logical partition stabilityVSAvoidI/O operation continuity

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The hypervisor detects server partition availability status in advance and proactively suspends client partition dispatch before I/O operations can fail. This preliminary action prevents the crash by stopping access attempts before they occur, rather than reacting after failure happens.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements a feedback mechanism where the hypervisor continuously monitors the availability status of the virtual I/O server partition and dynamically adjusts client partition dispatch based on this feedback. When the server becomes unavailable, the feedback triggers suspension of client access; when available again, it resumes access.

Inventive Principle:

Principle #23Feedback

2Reliability

If the hypervisor suspends dispatch to the client logical partition when the server is unavailable, then client partition crashes are prevented, but I/O operations are interrupted during server downtime

Engineering Contradiction:

Improveclient logical partition stabilityVSAvoidI/O operation delay

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The hypervisor performs preliminary detection of server availability and suspends client dispatch in advance, preventing the time loss that would occur from actual I/O failures and recovery attempts. This proactive approach trades controlled suspension for avoidance of catastrophic failure and longer recovery time.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system cushions against potential I/O failures by preemptively suspending client partition access when server unavailability is detected. This beforehand cushioning prevents the harmful effect of crashes and allows for smooth resumption when the server recovers, minimizing overall operational disruption.

Inventive Principle:

Principle #11Beforehand cushioning (Prior cushioning)

3Ease of repair

If the virtual I/O server partition is rebooted or fails, then server maintenance or recovery can proceed, but client logical partitions attempt to access unavailable physical storage causing crashes

Engineering Contradiction:

Improveserver recovery capabilityVSAvoidclient logical partition stability

Core Design Contradiction:

Ease of repairVSReliability

Solution Approach 1:

The hypervisor uses feedback from server partition status monitoring to dynamically control client partition access. When server reboot or failure is detected, the feedback mechanism triggers suspension of client dispatch, preventing crashes during recovery operations. When the server is back online, access is restored.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system takes preliminary action by detecting server unavailability and suspending client access before I/O operations can fail during server recovery. This prevents the cascade of failures that would occur if client partitions continued to access storage during server reboot or repair operations.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS9304802B2Mechanism for NPIV client recovery when NPIV server goes down

Publication Date: 2016.04.05 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

With N_Port ID Virtualization (NPIV), a managed system can be configured so that multiple logical partitions (LPARs) can access independent physical storage through the same physical fiber channel adapter. An NPIV client recovery component of a virtualization management component, such as a Power Hypervisor (pHYP), provides the emulation mapping between server and client virtual fiber channel adapters. The pHYP also provides a mechanism that prevents client partition crashes when the NPIV server (e.g., a VIOS logical partition) goes down. When the NPIV server is rebooted or powers down, the pHYP handles the client LPARs to avoid a crash by removing processing resources from the client logical partition. Thereby, the client logical partition is prevented from attempting to access a root volume group in physical storage via the NPIV server. The pHYP allocates processor resources to the client LPAR when the NPIV server is again available for I/O processing.