Method for determining system topology graph changes in a distributed computing system

Patent metadata

Field	Value
Publication number	US-9455881-B2
Application number	US-201314054011-A
Country	US
Kind code	B2
Filing date	Oct 15, 2013
Priority date	Oct 26, 2012
Publication date	Sep 27, 2016
Grant date	Sep 27, 2016

Abstract

A distributed computing system provides at least one service, and the service is provided by resources of the distributed computing system. Resources are represented by a topology graph including storage, application and database server nodes joined by edges indicative of relationships between nodes. Predefined parameters are monitored that are indicative of operating characteristics of the distributed computing system when providing the service. Values of the predefined parameters are compared with respective predetermined performance threshold values. The performance threshold values are determined based on a service level agreement (SLA). The SLA includes a service level objective (SLO). The performance threshold values are indicative of the respective operating characteristics prior to an event. Rules are provided for selecting a combination of application, storage and database server nodes to meet the SLO for each parameter of the parameter. Data associated with the set of rules are stored in the computing system.

First claim

What is claimed is:

1. A computer implemented method executed in a distributed computing system providing at least one service, wherein the at least one service is provided by resources of the distributed computing system, wherein the resources are represented by a topology graph comprising storage, application and database server nodes joined by edges indicative of relationships between nodes, wherein data associated with the topology graph being stored in the computing system, the method comprising:
receiving requests to provide the at least one service;
deploying the at least one service on the resources in accordance with the topology graph;
using the resources to provide the at least one service;
monitoring one or more predefined parameters indicative of respective one or more operating characteristics of the distributed computing system when providing the at least one service;
comparing values of the one or more predefined parameters with respective predetermined performance threshold values, wherein the performance threshold values are determined based on a service level agreement, SLA, wherein the SLA comprises at least one service level objective, SLO, the performance threshold values being indicative of the respective operating characteristics prior to an event;
providing a set of rules for selecting a combination of application, storage and database server nodes to meet the at least one SLO for each parameter of the one or more parameters, wherein data associated with the set of rules are stored in the computing system;
in response to a determination, at a time t0, that at least one parameter value of the one or more parameters values is higher or equal to the respective performance threshold value, using reference data and the set of rules to update the topology graph to prevent
occurrence of the event, the reference data being stored in association with the at least one service, wherein the reference data are determined by:
monitoring the one or more parameters during a time period, thereby providing monitoring data; and
using the monitoring data to generate the reference data, wherein the reference data comprises a table having one or more columns, each column representing a parameter of the one or more parameters, each column comprising a set of parameter values each associated with a time point of the time period at which the parameter has the parameter value; and
redeploying resources based on the updated topology graph.

2. The method according to claim 1 , wherein the distributed computing system is a cloud computing system.

3. The method according to claim 2 , wherein the at least one SLO comprises a condition specification identifying a condition that is selected from the group consisting of:
the average execution time of the at least one service per user is below a predetermined execution time threshold value; and
the CPU usage of the at least one service is below a preset maximum CPU usage value.

4. The method according to claim 3 , wherein the one or more predefined parameters comprises the execution time of the at least one service, wherein the performance threshold value associated with the execution time is equal to the execution time threshold value minus a predetermined execution time shift value.

5. The method according to claim 4 , wherein the reference data are determined using empirical data obtained by the same computing system at a previous time and/or using a simulation based on a model of the computing system.

6. The method according to claim 5 , wherein using the reference data comprises for each parameter of the one or more parameters:
reading the reference data;
determining a first time point of the time period corresponding to the performance threshold value associated with the parameter, wherein the first time point corresponds to the time t0;
generating from the reference data a subset of data containing the parameter values determined during the time between the first time point and the end time point of the time period.

7. The method according to claim 6 , wherein the updating comprises:
reading the subset of data;
determining a second time point of the time period corresponding to a second parameter value of the parameter;
comparing the performance threshold value with the second parameter value;
updating the topology graph based on the comparison result and the set of rules.

8. The method according to claim 6 , further comprising:
reading the subset of data;
producing a mathematical function representing the behavior of the variation of parameter values of the parameter;
using the mathematical function to determine a second time point of the time period and its corresponding second parameter value of the parameter;
comparing the performance threshold value with the second parameter value; and
updating the topology graph based on the comparison result and the set of rules.

9. The method according to claim 8 , further comprising:
in case the second parameter value is higher than the performance threshold value and the second parameter value is violating the at least one SLO, allocating additional resources to the at least one service, thereby updating the topology graph and the associated stored data in accordance with the set of rules.

10. The method according to claim 8 , further comprising:
in case the second parameter value is smaller than the performance threshold value, de-allocating resources to the at least one service, thereby updating the topology graph and the associated stored data in accordance with the set of rules.

11. The method according to claim 10 , wherein the at least one service comprises a first, a second and a third service, the method further comprising: for each service of the first, second and the third services executing the deploying, monitoring, comparing, providing, and update, using respective reference data, wherein the reference data are determined for each one of the first, second and the third services while the computing system provides the first, second and the third services at the same time; deploying the first, second and third services on resources in accordance with the first, second and third topology graph respectively; using the resources to provide the first, second and third services.

12. A non-transitory computer-readable medium executable in a distributed computing system providing at least one service, wherein the at least one service is provided by resources of the distributed computing system, wherein the resources are represented by a topology graph comprising storage, application and database server nodes joined by edges indicative of relationships between nodes, wherein data associated with the topology graph being stored in the computing system, the computer-readable medium comprising computer-readable program code embodied therewith which, when executed by a processor, cause the processor to execute a method, the method comprising:
receiving requests to provide the at least one service;
deploying the at least one service on the resources in accordance with the topology graph;
using the resources to provide the at least one service;
monitoring one or more predefined parameters indicative of respective one or more operating characteristics of the distributed computing system when providing the at least one service;
comparing values of the one or more predefined parameters with respective predetermined performance threshold values, wherein the performance threshold values are determined based on a service level agreement, SLA, wherein the SLA comprises at least one service level objective, SLO, the performance threshold values being indicative

Assignees

• IBM

Inventors

• Boerner Andreas
• Lebutsch David
• Mega Cataldo
• Zeng Chun Guang

Classifications

• H04L43/0876

  Network utilisation, e.g. volume of load or congestion level · CPC title

• G06F9/5027

  the resource being a machine, e.g. CPUs, Servers, Terminals · CPC title

• G06F9/5011

  the resources being hardware resources other than CPUs, Servers and Terminals · CPC title

• G06F9/5061

  Partitioning or combining of resources · CPC title

• H04L41/5096 Primary

  wherein the managed service relates to distributed or central networked applications · CPC title