US 9,432,256 B2Granted

Resource management method and resource management system

AssigneeHitachi, Ltd.

InventorsAtsumi Terayama, Toshio Otani

Life of this patent

Estimating the term…

Loading patent…

Abstract

In a cloud environment where multiple applications having different performance characteristics are operated in a mixture, in order to specify the cause of performance deterioration and to solve the confliction of requests related to a shared resource, the present invention detects the performance deterioration of resources allocated to multiple instances constituting a logical server configuration, and extracts an instance sharing resources existing performance deterioration, and compares a performance tendency of the extracted instance and a performance pattern showing a characteristics extracted from a performance information of the resource in association with a bottleneck factor, to thereby estimate the bottleneck factor by calculating mutual similarities and determining a performance pattern where the calculated similarity becomes maximum.

Claims6

Dependent claims

A resource management method targeting multiple instances having a logical server, the method comprising:

retaining one or more performance patterns that include a characteristic quantity extracted from a performance information of the resource in association with a bottleneck factor;
detecting deterioration of performance of one or more resources assigned to a first instance, where the one or more resources include a first virtual volume associated with the first instance;
extracting the first instance sharing a particular resource exhibiting the performance deterioration;
calculating a mutual similarity by comparing a performance tendency showing properties and characteristics of performance of the first instance and each of the one or more performance patterns;
estimating a bottleneck factor for a particular performance pattern from the one or more performance patterns having a maximum mutual similarity;
selecting an optimal candidate configuration from a plurality of potential configuration candidates based on the bottleneck factor, wherein each of the plurality of potential configuration candidates includes one or more candidate resources, and the one or more candidate resources include a second virtual volume;
and
automatically migrating the first virtual volume to the second virtual volume based on the optimal candidate configuration,
wherein:
one of the performance patterns is a characteristics of timing in which a threshold for determining deterioration of performance of the resource is exceeded in time series;
and
the similarity is calculated based on a correlation between one of the performance patterns and a time-series fluctuation of performance of at least one of the one or more resources of the first instance.

9 more elements

A resource management system comprising:

a performance pattern database, wherein the performance pattern database includes one or more performance patterns and each performance pattern includes a characteristic quantity extracted from a performance information of the resource in association with a bottleneck factor;
a resource configuration management unit configured to:
detect deterioration of performance of one or more resources assigned to a first instance, where the one or more resources include a first virtual volume associated with the first instance, and
extract the first instance of sharing a particular resource exhibiting the performance deterioration;
a mutual similarity calculating unit configured to:
calculate a mutual similarity by comparing a performance tendency showing properties and characteristics of performance of the first instance and each of the one or more performance patterns, and
estimate a bottleneck factor for a particular performance pattern from the one or more performance patterns having a maximum mutual similarity;
and
an optimal candidate configuration implementation unit configured to:
select an optimal candidate configuration from a plurality of potential configuration candidates based on the bottleneck factor, wherein each of the plurality of potential configuration candidates includes one or more candidate resources, and the one or more candidate resources include a second virtual volume;
and
automatically migrate the first virtual volume to the second virtual volume based on the optimal candidate configuration
wherein:
one of the performance patterns is a characteristics of timing in which a threshold for determining deterioration of performance of the resource is exceeded in time series;
and
the calculation of the similarity is based on an interrelationship between one of the performance patterns and a time-series fluctuation of performance of at least one of the one or more resources of the first instance.

13 more elements

Dependencies are inferred from the claim text.

Facts

Application: 14/425,525Filed 27 Mar 2014
Published: US 2015/0350019 A13 Dec 2015
Granted: 30 Aug 2016
Term ends: Not estimatedThe source records don’t support an estimate.

Classification17 codes

H04L 41/0816
Characterised by the conditions triggering a change of settings: the condition being an adaptation, e.g. in response to network events
G06F 9/50
Multiprogramming arrangements: allocation of resources, e.g. of the central processing unit [CPU]
G06F 9/5072
Partitioning or combining of resources: grid computing
G06F 11/3003
Monitoring: monitoring arrangements specially adapted to the computing system or computing system component being monitored