Micro-partition CPU Resource Allocation for High Availability Failover

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

High availability solutions for micro-partitions in computing systems require significant processing resources for failover support, leading to resource inefficiencies and performance overhead, especially when secondary logical partitions must match primary resources for failover.

Innovation Solution

Implementing a method where primary and secondary processing unit entitlements are determined for logical partitions, with the secondary entitlement being less than the primary, allowing for allocation and reservation of processing units from shared processor pools, and relocating the partition to a secondary pool in case of hardware failure, thereby reducing resource requirements.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If secondary logical partitions are configured with the same processing unit entitlements as primary partitions, then failover capability is ensured, but CPU resource consumption increases significantly

Engineering Contradiction:

Improvefailover capabilityVSAvoidCPU resource consumption

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent applies partial action by configuring secondary logical partitions with reduced processing unit entitlements (e.g., 0.5 PU) compared to primary partitions (e.g., 1.0 PU). This partial allocation is sufficient for failover scenarios where workloads are typically lighter, thereby reducing CPU resource consumption while maintaining adequate failover capability.

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The patent changes the parameter of processing unit entitlement from equal values in primary and secondary partitions to unequal values where secondary partitions have reduced entitlements. This parameter change enables resource optimization by matching the actual resource needs during failover scenarios, which typically involve resumed workloads rather than full production workloads.

Inventive Principle:

Principle #35Parameter changes

2Reliability

If secondary processing units are reserved equal to primary entitlements, then instantaneous failover is achieved, but shared processor pool resources are over-allocated

Engineering Contradiction:

Improvefailover speedVSAvoidshared processor pool resources

Core Design Contradiction:

ReliabilityVSQuantity of substance

Solution Approach 1:

The patent applies partial action by reserving only a portion of the processing units in the shared processor pool for secondary logical partitions. Instead of reserving full entitlements, the system reserves reduced amounts (e.g., 0.5 PU per secondary partition), which is sufficient for failover scenarios but prevents over-allocation of shared resources.

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The patent implements dynamic resource allocation where the actual processing units granted during failover can exceed the reserved secondary entitlements. The partition manager dynamically allocates additional processing units from the shared pool based on actual failover needs, allowing flexible resource utilization while maintaining reserve guarantees.

Inventive Principle:

Principle #15Dynamics

3Reliability

If all logical partitions fail over to secondary hosts simultaneously, then system continuity is maintained, but additional processing resources are required

Engineering Contradiction:

Improvesystem continuityVSAvoidprocessing power requirements

Core Design Contradiction:

ReliabilityVSPower

Solution Approach 1:

The patent applies partial action by configuring secondary hosts with reduced processing unit entitlements that are sufficient for failover scenarios. Since not all partitions need full processing power during failover (workloads are typically resumed rather than fully active), the reduced entitlements enable simultaneous failover without requiring additional processing resources.

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The patent implements preliminary action by pre-configuring secondary logical partitions with reduced processing unit entitlements and reserving corresponding resources in advance. This preliminary configuration ensures that when failover occurs, the necessary processing resources are already allocated and ready, enabling simultaneous failover of all partitions without requiring additional processing power at the moment of failover.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS9189381B2Managing CPU resources for high availability micro-partitions

Publication Date: 2015.11.17 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

A primary processing unit entitlement is determined for a logical partition. A smaller secondary processing unit entitlement is also determined. A partition manager allocates primary processing units to the logical partition from a primary shared processor pool, and the logical partition is activated. The secondary processing units are reserved for the logical partition from a secondary shared processor pool, and the logical partition can be relocated to the secondary shared processor pool in response to a condition such as a hardware failure. The logical partition can continue to process its workload with the fewer processor resources, and can be restored to the primary processing unit entitlement.