Predicting service delivery metrics using system performance data

Patent metadata

Field	Value
Publication number	US-9942103-B2
Application number	US-201314015007-A
Country	US
Kind code	B2
Filing date	Aug 30, 2013
Priority date	Aug 30, 2013
Publication date	Apr 10, 2018
Grant date	Apr 10, 2018

Abstract

A method for predicting a computerized service delivery organization workload including collecting data of a computer implementing service delivery routine including overlapping samples of load values, overlapping samples of performance values, overlapping samples of event outputs, ticket values and ticket volumes, building a first correlation of said load values with said performance values for predicting new performance values based on new data, building a second correlation of said performance values with said event outputs, said ticket values and said ticket volumes, combining said first and second correlations into a third correlation for correlating said load values with a ticket volume and an event volume, and determining at least one projected event volume or projected ticket volume value using said third correlation and at least one projected load value of said computer.

First claim

What is claimed is:

1. A method for configuring a computer system having a service delivery workload comprising:
receiving, by an automatic monitoring agent, a projected system load for a projected server configuration of said computer system, wherein a current server configuration of said computer system comprises a plurality of machines in an interdependent infrastructure forming a queuing system and wherein said projected server configuration differs from said current configuration;
estimating, by said automatic monitoring agent, a load value of said projected system load and said projected server configuration of said computer system, wherein said load value is determined given a plurality of ticket generation policies describing rules of said automated monitoring agent generating system alert events and system alert tickets, wherein said automatic monitoring agent monitors said current server configuration of said computer system and generates a plurality of system alert tickets dependent on actual load values of said computer system and actual performance values of said computer system, wherein system alert ticket values and system alert ticket volumes are determined from said plurality of system alert tickets;
determining, by said automatic monitoring agent, that said load value of said projected server configuration correlates to said actual performance data of said current server configuration of said computer system;
estimating, by said automatic monitoring agent, said workload of said projected server configuration of said computer system using a correlation model correlating said load value with said system alert ticket volume and said system alert ticket volumes; and
adding, by said automatic monitoring agent, at least one machine to said current server configuration of said computer system according to an infrastructure defined by said projected server configuration of said computer system given said workload estimated using said correlation model.

2. The method of claim 1 , further comprising collecting, by said automatic monitoring agent, data of said current server configuration of said computer system, said data including overlapping samples of load values, overlapping samples of performance values, overlapping samples of system alert event outputs, system alert ticket values and system alert ticket volumes, wherein collecting the data further comprises collecting configuration values of said current server configuration of said computer system and connectivity in said interdependent infrastructure forming a queuing system.

3. The method of claim 2 , further comprising pruning said data to obtain a sub-set of said data.

4. The method of claim 2 , further comprising pruning said data to remove at least one of said overlapping samples failing outside of a domain of performance values associated with ticket data.

5. The method of claim 2 , further comprising pruning said data to remove at least one of said ticket values determined to be unrelated to said workload given an expected time service said at least one of said ticket values.

6. The method of claim 2 , further comprising:
building, by said automatic monitoring agent, a load-to-performance correlation of said load value of said projected server configuration with said actual performance values of said computer system; and
building, by said automatic monitoring agent, a performance-to-workload correlation of said actual performance values of said computer system with said event outputs, said ticket values and said ticket volumes.

7. A computer program product for configuring a computer system having a service delivery workload comprising:
a non-transitory computer readable storage medium having computer readable program code embodied therewith, the computer readable program code comprising:
computer readable program code configured to receive a projected system load for a projected server configuration of said computer system, wherein a current server configuration of said computer system comprises a plurality of machines in an interdependent infrastructure forming a queuing system and wherein said projected server configuration differs from said current configuration;
computer readable program code configured to estimate a load value of said projected system load and said projected server configuration of said computer system, wherein said load value is determined given a plurality of ticket generation policies describing rules of an automated monitoring agent generating system alert events and system alert tickets, wherein said automatic monitoring agent monitors said current server configuration of said computer system and generates a plurality of system alert tickets dependent on actual load values of said computer system and actual performance values of said computer system, wherein system alert ticket values and system alert ticket volumes are determined from said plurality of system alert tickets;
computer readable program code configured to determine that said load value of said projected server configuration load correlates to said actual performance data of said current server configuration of said computer system;
computer readable program code configured to estimate said workload of said projected server configuration of said computer system using a correlation model correlating said load value with said system alert ticket volume and said system alert ticket volumes; and
computer readable program code configured to add at least one machine to said current server configuration of said computer system according to an infrastructure defined by said projected server configuration of said computer system given said workload estimated using said correlation model.

8. The computer program product of claim 7 , further comprising computer readable program code configured to collect data of said current server configuration of said computer system, said data including overlapping samples of load values, overlapping samples of performance values, overlapping samples of system alert event outputs, system alert ticket values and system alert ticket volumes, wherein collecting the data further comprises collecting configuration values of said current server configuration of said computer system and connectivity in said interdependent infrastructure forming a queuing system.

9. The computer program product of claim 8 , further comprising computer readable program code configured to prune said data to obtain a sub-set of said data.

10. The computer program product of claim 8 , further comprising computer readable program code configured to prune said data to remove at least one of said overlapping samples failing outside of a domain of performance values associated with ticket data.

11. The computer program product of claim 8 , further comprising computer readable program code configured to prune said data to remove at least one of said ticket values determined to be unrelated to said workload given an expected time service said at least one of said ticket values.

12. The computer program product of claim 8 , further comprising:
computer readable program code configured to build a load-to-performance correlation of said load value of said projected server configuration with said actual performance values of said computer system; and
computer readable program code configured to build a performance-to-workload correlation of said actual performance values of said computer system with said event outputs, said ticket values and said ticket volumes.

Assignees

• IBM

Inventors

• Branch Joel W
• Diao Yixin
• Olsson Emi K
• Paradkar Amitkumar M
• Puri Rajeev
• Shwartz Larisa

Classifications

• H04L41/147 Primary

  for predicting network behaviour · CPC title

• H04L41/5038

  Electricity · mapped topic

• H04L41/5074

  Handling of user complaints or trouble tickets · CPC title

• H04L43/55 Primary

  Testing of service level quality, e.g. simulating service usage · CPC title