Computer program, method, and information processing apparatus for analyzing performance of computer system

Patent metadata

Field	Value
Publication number	US-8984125-B2
Application number	US-201213587544-A
Country	US
Kind code	B2
Filing date	Aug 16, 2012
Priority date	Aug 16, 2012
Publication date	Mar 17, 2015
Grant date	Mar 17, 2015

Abstract

In an information processing apparatus, a comparing unit determines whether the response time of each transaction falls within an acceptable time range that is specified previously. For each time window, a first calculation unit calculates a load of processes executed in parallel by the servers in a specified tier, based on transaction data of individual transactions. Further, a second calculation unit calculates a total progress quantity in each time window, based on the transaction data of transactions whose respective response times are determined to fall within the acceptable time range. A determination unit determines a specific load value as a threshold at which the total progress quantity begins to decrease in spite of an increase of the load.

First claim

What is claimed is:

1. A computer-readable non-transitory recording medium storing therein an analyzing program, that causes a computer to execute a process comprising:
determining whether a response time of each transaction falls within an acceptable time range that is specified previously, based on transaction data that records a plurality of transactions executed in a coordinated way by a plurality of servers in a multi-tier system during a given analysis period, the transaction data indicating execution periods of processes executed by the servers in each tier of the multi-tier system;
dividing the given analysis period into a plurality of time windows and calculating, for each window, a load of the processes executed in parallel by the servers in a specified tier, based on the transaction data of the individual transactions;
calculating, based on the transaction data of the transactions whose response times are determined to fall within the acceptable time range, a total throughput in each time window by adding up throughputs of the processes executed, as part of the transactions whose response times are determined to fall within the acceptable time range, by the servers in the specified tier during said each time window, a throughput of a process executed in a time window being calculated as a ratio of a portion of the process that is executed in the time window to the entire process invoked in response to a processing request therefor, the ratio being weighted according to a processing time, in low workload conditions, of a process category to which the executed process belongs; and
determining, based on the calculated load and total throughput of each time window, a first load value as a first threshold at which the total throughput begins to decrease in spite of an increase of the load.

2. The computer-readable non-transitory recording medium according to claim 1 , the procedure further comprising:
determining, based on the calculated load and total throughput of each time window, a second load value as a second threshold at which a ratio of an increase of the total throughput to an increase of the load begins to fall below a specified value; and
configuring the servers in the specified tier with an allowable limit of concurrent processes, such that the load of the servers falls within a range delimited by the first threshold and second threshold even when the servers are loaded with as many concurrent processes as the allowable limit.

3. The computer-readable non-transitory recording medium according to claim 1 , wherein the determining of the first threshold comprises:
obtaining a minimum value and a maximum value among the loads in the time windows, and dividing a range between the minimum value and the maximum value into a plurality of subranges,
calculating an average throughput in each subrange by averaging the total throughputs in the time windows whose respective loads fall within said each subrange, and
comparing the average throughputs between the plurality of subranges to determine at which load value the total throughput begins to decrease in spite of an increase of the load.

4. The computer-readable non-transitory recording medium according to claim 3 , wherein:
the procedure further comprises determining, based on the calculated load and total throughput of each time window, a second load value as a second threshold at which a ratio of an increase of the total throughput to an increase of the load begins to fall below a specified value; and
the determining of the first threshold further comprises:
selecting one of the subranges which contains the first load value determined as the second threshold,
detecting one of the subranges whose average throughput has a statistically significant difference from the average throughput in the selected subrange, and
determining the first threshold from among load values in the detected subrange.

5. The computer-readable non-transitory recording medium according to claim 4 , wherein the detecting of one of the subranges uses t-tests to determine whether there is a statistically significant difference in the average throughputs.

6. The computer-readable non-transitory recording medium according to claim 1 , wherein the specified tier in the multi-tier system is a tier accommodating one or more servers that have reached performance limit thereof.

7. An analyzing method executed by a computer, the analyzing method comprising:
determining whether a response time of each transaction falls within an acceptable time range that is specified previously, based on transaction data that records a plurality of transactions executed in a coordinated way by a plurality of servers in a multi-tier system during a given analysis period, the transaction data indicating execution periods of processes executed by the servers in each tier of the multi-tier system;
dividing the given analysis period into a plurality of time windows and calculating, for each window, a load of the processes executed in parallel by the servers in a specified tier, based on the transaction data of the individual transactions;
calculating, based on the transaction data of the transactions whose response times are determined to fall within the acceptable time range, a total throughput in each time window by adding up throughputs of the processes executed, as part of the transactions whose response times are determined to fall within the acceptable time range, by the servers in the specified tier during said each time window, a throughput of a process executed in a time window being calculated as a ratio of a portion of the process that is executed in the time window to the entire process invoked in response to a processing request therefor, the ratio being weighted according to a processing time, in low workload conditions, of a process category to which the executed process belongs; and
determining, based on the calculated load and total throughput of each time window, a first load value as a first threshold at which the total throughput begins to decrease in spite of an increase of the load.

8. An information processing apparatus, comprising:
a memory configured to store transaction data that records a plurality of transactions executed in a coordinated way by a plurality of servers in a multi-tier system during a given analysis period, the transaction data indicating execution periods of processes executed by the servers in each tier of the multi-tier system; and
a processor configured to perform a procedure including:
determining whether a response time of each transaction falls within an acceptable time range that is specified previously, based on the transaction data
dividing the given analysis period into a plurality of time windows and calculating, for each window, a load of the processes executed in parallel by the servers in a specified tier, based on the transaction data of the individual transactions;
calculating, based on the transaction data of the transactions whose response times are determined to fall within the acceptable time range, a total throughput in each time window by adding up throughputs of the processes executed, as part of the transactions whose response times are determined to fall within the acceptable time range, by the servers in the specified tier during said each time window, a throughput of a process executed in a time window being calculated as a ratio of a portion of the process that is executed in the time window to the entire process invoked in response to a processing request therefor, the ratio being weighted according to a processing time, in low workload conditions, of a process category to which the executed process belongs; and
determining, based on the calculated load and total thro

Assignees

• Kanemasa Yasuhiko
• Pu Calton
• Wang Qingyang
• Fujitsu Ltd
• Georgia Tech Res Inst

Inventors

• Kanemasa Yasuhiko
• Pu Calton
• Wang Qingyang

Classifications

• G06F11/3419

  by assessing time · CPC title

• G06F2201/81

  Threshold · CPC title

• G06F11/3433

  for load management (allocation of a server based on load conditions G06F9/505; load rebalancing G06F9/5083; redistributing the load in a network by a load balancer H04L67/1029) · CPC title

• H04L67/1029 Primary

  using data related to the state of servers by a load balancer · CPC title

• G06F11/34

  Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation {; Recording or statistical evaluation of user activity, e.g. usability assessment} · CPC title