JP5006837B2 - USAGE PREDICTION SYSTEM, USAGE PREDICTION METHOD AND USAGE PREDICTION PROGRAM - Google Patents

Description

  The present invention relates to a usage amount prediction system, a usage amount prediction method, and a usage amount prediction program for predicting an increase in usage amount of a migration destination system when a system is migrated.

  Today, various databases are constructed using a large-capacity data storage device or the like. For example, in a company or the like, business data that is successively generated is stored in such a database. In addition, a plurality of departments or users may share the data storage device in managing the large-capacity data storage device. In this case, the memory capacity may be allocated and managed based on the used size acquired from the manager of each database. In this case, when the usage amount approaches the allocated amount, an inquiry is made to the administrator. However, if the administrator is frequently inquired, the response of the administrator becomes complicated. Also, some managers may apply for an oversized size. In this case, unnecessary memory areas that are not used increase, and the use efficiency of the data storage device may be lowered. Therefore, a technique for efficiently managing a database device shared by a plurality of users has been studied (for example, see Patent Document 1).

In this document, the usage amount of the database device is acquired, and the usage amount is registered in the usage history data storage unit. When the used capacity exceeds the alarm reference value, the management computer issues an alarm notification. Then, the management computer calculates an average value of this user's usage amount in the past one month in the database device and an average value of the usage amount one year ago, and based on the increment of these average values, a new allocation amount is calculated. Set up.
Japanese Patent Laying-Open No. 2007-213339 (FIG. 4)

  By the way, a system that has been used up to now may be migrated to another new system. In this case, if the capacity of the memory and CPU constituting the new system is not sufficient, the processing capacity in the new system will be lower than that in the old system. Further, when a large margin is provided for the memory and the CPU, the utilization efficiency is deteriorated. Therefore, it is desirable to predict the amount of memory and CPU used in the new system, and to identify the optimum memory and CPU from this predicted amount of usage. In this case, it is difficult to predict an appropriate usage amount when the usage amount of the memory or CPU in the old system is unknown.

  The present invention has been made in view of the above-described problems, and its purpose is to use the migration destination system after migration if the number of logins of the migration source system is known even if the usage amount of the migration source system is unknown. Is to provide a usage amount prediction system, a usage amount prediction method, and a usage amount prediction program.

In order to solve the above-described problem, the invention described in claim 1 includes a migration source login number data storage unit storing a login date and a logout date and time of a migration source user using the migration source system, and the migration source system. The migration destination login number data storage means storing the login date and logout date and time of the migration destination user using the migration destination system used together, and the actual usage amount actually used in the server constituting the migration destination system The usage amount data storage means to be recorded together with the migration source system and the migration destination system are used together so that all users using the migration source system use the migration destination system. the amount and control means for predicting the amount of the target system when it becomes the transition state to the predicted A Temu, said control means from said migration source logins data storage means to identify the number of the source user at a particular time and date, the number of target users in the specific time from the destination log number of data storage means identified, obtains the number of the source user who the specific, the scale factor calculation means for calculating the scale factor divided by the specified destination number of users, the usage record from the usage result data storage unit, the The gist of the invention is that the apparatus includes a prediction unit that calculates the usage amount of the migration destination system using a value obtained by multiplying the actual usage amount and the calculated scale factor.

  The invention according to claim 2 is the usage amount prediction system according to claim 1, wherein the prediction unit subtracts the fixed usage amount from the usage amount result acquired from the usage amount result data storage unit, The gist is to calculate an increase load amount by multiplying by the scale factor, and to calculate the use amount by adding the actual use amount to the increase load amount.

  According to a third aspect of the present invention, in the usage amount prediction system according to the first or second aspect, the control unit obtains the number of migration source users from the migration source login number data storage unit, and the migration destination login. Peak number specifying means for acquiring the number of migration destination users from number data storage means, summing up the number of migration source users and the number of migration destination users for each day, and identifying a peak day with the largest number of total logins; And calculating the sum of the number of migration source users and the number of migration destination users on the peak date, and further comprising a peak time specifying means for specifying the time with the largest total number of users as a peak time, and the scale factor The calculating means acquires the number of migration source users and the number of migration destination users at the identified peak time, and uses the number of migration source users and the number of migration destination users to regulate Calculating a coefficient, said prediction means, and summarized in that for calculating the amount by using the usage record in the peak day.

  According to a fourth aspect of the present invention, in the usage amount prediction system according to any one of the first to third aspects, the migration destination system includes a plurality of servers, and the control means has a normal operating time. The limit usage amount of the band and the server is stored, the usage amount data storage unit stores data related to the usage amount for each server, and the prediction unit uses the usage amount for each server. The actual usage amount is obtained from the actual usage amount data storage unit, and the usage amount is calculated for each server using the actual usage amount, and the control unit determines the predicted usage amount in the normal operation time zone as the limit of the server. Means for comparing with the usage amount, means for specifying the server associated with the predicted usage amount as another distribution source server when the predicted usage amount is greater than or equal to the limit usage amount, and a distribution source server In one case, the excess usage amount is calculated by subtracting the limit usage amount from the maximum predicted usage amount of the distribution source server. If there are multiple distribution source servers, the maximum predicted usage amount of each distribution source server is calculated. Subtracting the limit usage from the value, summing them up at each time, and specifying the maximum predicted usage total as the excess usage, and the capacity that can be distributed by subtracting the fixed usage from the limit usage Means for calculating the excess usage amount calculated by dividing the calculated amount of excess use by the amount that can be distributed, and determining and recording the number of additional servers, and a limit from the maximum predicted usage amount of each distribution source server Means for dividing the value obtained by subtracting the usage amount by the limit usage amount, and calculating a load allocation rate in the distribution source server; and means for outputting the recorded number of additional servers and the load allocation rate. This is the summary.

The invention according to claim 5 uses the migration source login number data storage means storing the login date and logout date and time of the migration source user using the migration source system, and the migration destination system used in combination with the migration source system. Migration destination login number data storage means storing the login date and logout date and time of the migration destination user , and usage amount data storage means for recording the actual usage amount actually used in the server constituting the migration destination system together with the time And when the migration source system and the migration destination system are used together, all users using the migration source system are migrated to use the migration destination system. of the target system predicted usage with usage prediction system and a control means for predicting the amount of A Nau method, said control means from said migration source logins data storage means to identify the number of the source user at a particular time, the target number of login data stores the number of target users in the specific date identify the unit, the number of the source user identified, the scale factor calculation step of calculating a scale factor divided by the specified destination number of users, and obtains the usage record from the usage result data storage unit The gist is to execute a prediction step of calculating the usage amount of the migration destination system using a value obtained by multiplying the actual usage amount and the calculated scale factor.

The invention according to claim 6 uses the migration source login number data storage means storing the login date and logout date and time of the migration source user using the migration source system, and the migration destination system used in combination with the migration source system. Migration destination login number data storage means storing the login date and logout date and time of the migration destination user , and usage amount data storage means for recording the actual usage amount actually used in the server constituting the migration destination system together with the time And when the migration source system and the migration destination system are used together, all users using the migration source system are migrated to use the migration destination system. of the target system predicted usage with usage prediction system and a control means for predicting the amount of A Nau program, said control means, said from the source log number data storing means, to determine the number of the source user at a particular time, the target number of login data stores the number of target users in the specific date identify the unit, the number of the source user specified scale factor calculation means for calculating the scale factor divided by the target number of users specified, and acquires the usage record from the usage result data storage unit The gist of the invention is to use the value obtained by multiplying the actual usage amount and the calculated scale factor to function as a predicting means for calculating the usage amount of the migration destination system.

(Function)
According to the present invention, the control means calculates a scale factor obtained by dividing the number of migration source users by the number of migration destination users at the same time. The control means uses the migration destination system usage amount when migrating the migration source user to the migration destination system using a value obtained by multiplying the usage quantity history obtained from the usage quantity history data storage means and the calculated scale factor. Is calculated. For this reason, the usage amount of the migration destination system after the migration can be predicted using the number of users of the migration source system and the migration destination system and the usage amount record of the migration destination system. Therefore, even if the usage amount of the migration source system is unknown, if the number of users is known, the usage amount of the migration destination system after migration can be predicted based on the number of users.

  According to the present invention, the control means calculates the increased load amount by multiplying the value obtained by subtracting the fixed usage amount from the usage amount result acquired from the usage amount result data storage means by the scale factor, and the increased load amount. The usage amount is calculated by adding the actual usage amount to. For this reason, since the increased load amount is calculated by multiplying the value obtained by subtracting the fixed use amount that does not depend on the number of users from the use amount, the total load amount of the migration destination system can be predicted more accurately. it can.

  According to the present invention, the control unit totals the number of migration source users acquired from the migration source login data storage unit and the number of migration destination users acquired from the migration destination login data storage unit every day, and the total number of users Identify the peak days with the most. The control means calculates the sum of the number of migration source users and the number of migration destination users on this peak day, and identifies the time with the largest total number of users as the peak time. The control means acquires the number of migration source users and the number of migration destination users at the specified peak time, calculates a scale factor using the number of migration source users and the number of migration destination users, and uses the actual usage on the peak day. To calculate usage. For this reason, the usage amount on the peak day with the largest number of users using the system is acquired, and the usage amount after the migration is specified. Further, on the peak day, the scale factor is calculated using the number of migration source users and the number of migration destination users at the time when the sum of the number of migration source users and the number of migration destination users is the largest. Therefore, a predicted value close to the maximum usage amount of the migration destination system after the migration can be calculated by using the actual usage amount when the usage amount becomes the largest. Furthermore, by calculating the scale factor using the number of users when the ratio of the number of one login to the whole is as small as possible, a more accurate usage amount can be calculated.

  According to the present invention, the control means compares the predicted usage amount in the normal operation time zone with the limit usage amount of the server. When the predicted usage is greater than or equal to the limit usage, the control means identifies the server associated with the predicted usage as a distribution source server that distributes the server to others. When there is one distribution source server, the control means calculates an excess usage amount obtained by subtracting the limit usage amount from the maximum predicted usage amount of the distribution source server. When there are a plurality of distribution source servers, the control means subtracts the limit use amount from the maximum value of the predicted use amount of each distribution source server, and sums them up for each time, and the total predicted use amount thus obtained is calculated. Identified as excess usage. The control means calculates a distributable underwriting amount obtained by subtracting the fixed usage amount from the limit usage amount. The control means divides the calculated excess usage amount by the available distribution amount to determine and record the number of additional servers. The control means calculates a burden allocation rate in the distribution source server by dividing a value obtained by subtracting the limit usage amount from the maximum predicted usage amount of each distribution source server by the limit usage amount. The control means outputs the recorded number of additional servers and the burden allocation rate. For this reason, when the predicted usage exceeds the limit usage during the normal operating hours, the control means automatically calculates and outputs the number of additional servers and the load allocation ratio that will fit within the server limit usage. To do. Therefore, the administrator can easily grasp the necessary additional server and its load allocation rate when the predicted usage amount of the migration destination system after migration is large.

  According to the present invention, the usage amount after system migration can be predicted.

  Hereinafter, an embodiment embodying the present invention will be described with reference to FIGS. In this embodiment, a load prediction system for predicting the load of the migration destination system will be described. In this embodiment, load prediction is performed when all users using the migration source system are migrated to the migration destination system from the current situation where the migration source system and the migration destination system are used together.

  In the present embodiment, the migration source system 20 as an old system and the migration destination system 30 as a new system are connected to a plurality of client terminals 10 via networks, respectively. The client terminal 10 instructs the migration source system 20 or the migration destination system 30 to execute processing. For this reason, the client terminal 10 includes an operation unit (keyboard and pointing device) and a display unit (display).

  The migration source system 20 includes a log data storage unit 21 as a migration source login number data storage unit. Log data is recorded in the log data storage unit 21. This log data is recorded when a login request is received from the client terminal 10. This log data includes data related to the login user identifier, the login date and time, and the logout date and time.

In the login user identifier data area, data relating to an identifier (user identifier) for specifying a user who has logged into the migration source system 20 is recorded.
In the login date / time data area and the logout date / time data area, data related to the date and time when the user logged in and logged out is recorded.

  On the other hand, the migration destination system 30 is a newly constructed computer system and includes three processing servers. In the present embodiment, it is assumed that these processing servers include a CPU having the same function and a memory having the same physical memory amount (for example, 3000 MB). In the present embodiment, each memory is a target for predicting the usage (load). Of these, the first server functions as a gateway and a report server. Furthermore, the second and third servers function as report servers. Since the first server also functions as a gateway, the resource allocation amount is set lower than that of the second and third servers.

  In this embodiment, in the migration destination system 30, it has already been used by some users, and its usage record is recorded. Therefore, the migration destination system 30 includes a log data storage unit 31 as a migration destination login number data storage unit and a usage amount result data storage unit 32 as a usage amount result data storage unit.

  Log data is recorded in the log data storage unit 31. This log data has the same data configuration as the log data recorded in the log data storage unit 21 described above. This log data is recorded when a login request is received from the client terminal 10. This log data includes data related to the login user identifier, the login date and time, and the logout date and time.

In the login user identifier data area, data relating to an identifier (user identifier) for specifying a user who has logged into the migration destination system 30 is recorded.
In the login date / time data area and the logout date / time data area, data related to the date and time when the user logged in and logged out is recorded.

  On the other hand, as shown in FIG. 2A, usage amount result data 320 relating to the memory usage amount (actual result) in the migration destination system 30 is recorded in the usage amount result data storage unit 32. The actual usage data 320 is recorded when the measured memory usage data is acquired. The usage amount actual data 320 includes data relating to the server identifier, date and time, total memory usage, and application memory usage.

Data relating to an identifier (server identifier) for specifying the server is recorded in the server identifier data area.
In the date / time data area, data relating to the date / time (year / month / day and time) of the memory usage of the server is recorded.

In the total memory usage data area, data related to the total memory usage used in this server is recorded.
In the application memory usage data area, data relating to the memory usage used for processing of applications used in the server is recorded. In this embodiment, the amount of memory used for executing an application based on Java (registered trademark) is recorded in this data area.

  The migration source system 20 and the migration destination system 30 described above are connected to the load prediction server 50. The load prediction server 50 executes a load prediction process when all users using the migration source system 20 use the migration destination system 30. The load prediction server 50 includes a control unit 51 as a control unit, a usage record data storage unit 52, a predicted load value data storage unit 53, and a simulation result data storage unit 54. The control unit 51 includes a CPU, a RAM, a ROM, and the like (not shown), and performs processing described later (processing including a peak day identification stage, a peak time identification stage, a scale factor calculation stage, a prediction stage, etc.). And the control part 51 functions as basic information acquisition means 511, load predicted value calculation means 512, output means 513, simulation condition acquisition means 514, simulation execution means 515, etc. by executing the usage amount prediction program for this purpose. To do.

The basic information acquisition unit 511 acquires basic information for calculating a predicted load value, and functions as a peak date specifying unit and a peak time specifying unit. The basic information acquisition unit 511 of the present embodiment executes processing for acquiring information for specifying the migration source system and the migration destination system. Further, when the migration source system and the migration destination system (20, 30) are designated, the basic information acquisition unit 511 obtains the number of logged-in users from these log data storage units (21, 31), and further, the migration destination system. The usage record is obtained from the 30 usage record data storage unit 30.

  The predicted load value calculation unit 512 functions as a scale factor calculation unit and a prediction unit, and predicts the load (usage amount) of the migration destination system 30 when the user of the migration source system 20 has migrated to the migration destination system 30. Execute. In this embodiment, the memory usage of each server is predicted. Therefore, the predicted load value calculation unit 512 stores a predicted load value calculation formula and a scale factor calculation formula used therefor. In the present embodiment, equation (1) in FIG. 4 is used as the scale factor calculation equation. Further, the formula (2) is used in FIG.

  Furthermore, the predicted load value calculation unit 512 stores a fixed usage amount calculation formula for calculating a fixed usage amount in the predicted load value calculation formula. In the present embodiment, the fixed usage calculation formula is a formula for adding the application memory usage and the OS memory usage. In addition, the predicted load value calculation unit 512 of this embodiment stores a certain amount of OS memory usage (for example, 500 MB).

The output unit 513 executes processing for outputting the calculated predicted usage amount or simulation result.
The simulation condition acquisition unit 514 executes processing for acquiring simulation conditions.

  The simulation execution means 515 executes a simulation process for allocating a predicted usage amount exceeding the physical memory amount to another server. For this purpose, the simulation execution means 515 stores a load allocable amount calculation formula for calculating a usage amount that can distribute the load to other servers. This load increase amount calculation formula is a formula for calculating a value excluding the fixed usage amount (total value of application memory usage amount and OS memory usage amount) in each server from the predicted usage amount calculated by the equation (2). It is.

  On the other hand, the usage amount data storage unit 52 records data on the memory usage (actual) used for calculating the predicted usage. This usage amount result data is acquired from the usage amount result data storage unit 32 of the migration destination system 30 and recorded when performing load prediction. This usage amount actual data has the same data configuration as the usage amount actual data recorded in the usage amount actual data storage unit 32. This usage amount actual data includes data relating to the server identifier, date and time, total memory usage, and application memory usage.

Data relating to an identifier (server identifier) for specifying the server is recorded in the server identifier data area.
In the date / time data area, data relating to the date / time (year / month / day and time) of the memory usage of the server is recorded.

In the total memory usage data area, data related to the total memory usage used in this server is recorded.
In the application memory usage data area, data relating to the memory usage used for processing of applications used in the server is recorded.

  In addition, as shown in FIG. 2B, predicted load value data 530 relating to the predicted load in the migration source system 20 is recorded in the predicted load value data storage unit 53. This predicted load value data 530 is recorded when a predicted load value for each server and each time is calculated. The predicted load value data 530 includes data relating to the server identifier, time, and predicted usage.

In the server identifier data area, data relating to an identifier (server identifier) for specifying a server that has predicted the load is recorded.
In the time data area, data relating to the time when the load is predicted is recorded.
Data on the predicted load (memory usage) is recorded in the predicted usage data area.

  The simulation result data storage unit 54 shown in FIG. 1 records data related to the simulation result regarding the distribution of the processing amount of each server using the predicted load value. This simulation result data is recorded when the simulation is executed. The simulation result data includes data relating to the server identifier, time, and simulation result predicted usage.

In the server identifier data area, data relating to an identifier (server identifier) for specifying a server whose load is predicted by simulation is recorded.
In the time data area, data relating to the time when the load is predicted is recorded.
In the simulation result predicted usage data area, data related to the load (memory usage) predicted by the simulation is recorded.

  Further, an input / output terminal 60 is connected to the load prediction server 50. The input / output terminal 60 uses the operation unit (keyboard or pointing device) to transmit data regarding conditions such as basic information acquisition means for the migration source system 20 and the migration destination system 30 and simulation resource allocation to the load prediction server 50. To do. In addition, the input / output terminal displays, on the display, the predicted use amount calculated and the predicted use amount of the memory of each server according to the resource allocation condition of the simulation using the load prediction server 50.

  A processing procedure of load prediction processing for predicting the load of the migration destination system 30 using the load prediction server 50 described above will be described with reference to FIG. Here, the memory usage of the first to third servers constituting the migration destination system 30 is predicted.

(Load prediction process)
First, the control unit 51 of the load prediction server 50 executes basic information acquisition processing (step S1-1). Specifically, the basic information acquisition unit 511 of the control unit 51 displays a load prediction information acquisition screen on the input / output terminal 60. This load prediction information acquisition screen is provided with an input field for specifying a migration destination system for performing load prediction, a migration source system, and a load prediction execution button.

  Here, the administrator inputs information for specifying the migration destination system 30 that performs load prediction and the migration source system 20, and selects a load prediction button. Thereby, the control unit 51 of the load prediction server 50 identifies the migration source system 20 and the migration destination system 30.

  Next, the control part 51 performs the acquisition process of the whole memory usage (step S1-2). Here, the control unit 51 specifies the reference date (in this embodiment, the peak date with the largest number of logged-in users in the most recent month). In this case, the control unit 51 obtains log data for a predetermined period (here, the latest one month) from the log data storage unit 21 of the identified migration source system 20, and counts and calculates the number of login users per day. To do. Specifically, the basic information acquisition unit 511 of the control unit 51 counts and calculates the number of logged-in users per day within the past month from the log-in date / time and log-out date / time of the log data in the log data storage unit 21.

Further, the control unit 51 acquires log data for a predetermined period (here, the last one month) from the log data storage unit 21 of the identified migration source system 20, and counts and calculates the number of login users per day. . Specifically, the basic information acquisition unit 511 of the control unit 51 counts and calculates the number of logged-in users per day within the past month from the log-in date / time and log-out date / time of the log data in the log data storage unit 21.

  Then, the basic information acquisition unit 511 of the control unit 51 sums the calculated number of login users of the migration source system 20 and the calculated number of login users of the migration destination system 30 for each date. The basic information acquisition unit 511 compares the total number of logged-in users, and identifies the day with the largest number of logged-in users as the peak day. The basic information acquisition unit 511 acquires the usage data 320 on the peak date from the usage data storage 32 and records it in the usage data storage 52.

  Next, the control unit 51 executes peak time specifying processing (step S1-3). Here, the control part 51 specifies the peak time of a peak day. Specifically, the basic information acquisition unit 511 of the control unit 51 extracts the log data of the peak date of the migration source system 20 from the log data storage unit 21. The basic information acquisition unit 511 calculates the log-in time transition by counting the log-in user number of the migration source system 20 for each peak day time from the log-in date and log-out date and time of the extracted log data. Further, the basic information acquisition unit 511 extracts the log data of the peak date of the migration destination system 30 from the log data storage unit 31. The basic information acquisition unit 511 calculates the log-in time transition by counting the log-in user number of the migration destination system 30 for each peak day time from the log-in date and log-out date and time of the extracted log data. Then, the basic information acquisition unit 511 calculates the total number time transition by adding the number of login users of the migration source system 20 and the number of login users of the migration destination system 30 for each same “time”. The basic information acquisition unit 511 specifies the time (peak time) with the largest number of logged-in users in the total number time transition. For example, as shown in FIG. 3, when the number of logged-in users at 9 o'clock is the largest, 9 o'clock is specified as the peak time (P1).

  Next, the control part 51 performs the calculation process of a scale factor (step S1-4). Specifically, the predicted load value calculation unit 512 of the control unit 51 substitutes the number of login users of the migration source system 20 at the peak time and the number of login users of the migration destination system 30 at the same time into the equation (1). To calculate the scale factor.

Next, the control part 51 performs the calculation process of a load predicted value (step S1-5). Here, the control unit 51 repeats the calculation of the fixed usage amount and the calculation of the predicted usage amount for each time.
First, the predicted load calculation unit 512 of the control unit 51 calculates a fixed usage amount using a fixed usage amount calculation formula. Specifically, the predicted load value calculation unit 512 sequentially extracts the usage amount result data recorded in the usage amount result data storage unit 52. Then, the load predicted value calculation unit 512 calculates the fixed usage by adding the OS memory usage to the fixed usage of the extracted usage data.

  Next, the predicted load value calculation unit 512 of the control unit 51 calculates the predicted usage amount at this time using a predicted load value calculation formula. Specifically, the predicted load calculation unit 512 calculates the predicted usage by substituting the total usage of the extracted usage data into the memory, the calculated fixed usage and the scale factor into the equation (2). To do. Then, the predicted load value calculation unit 512 records the calculated predicted usage amount in the predicted load value data storage unit 53 as the predicted load value data 530 together with the server identifier and the time of the extracted usage amount record data. The above process is executed for all of the usage record data recorded in the usage record data storage unit 52.

Next, the control part 51 performs the output process of a load predicted value (step S1-6). Specifically, the output unit 513 of the control unit 51 generates predicted value result display data including the predicted usage for each server identifier recorded in the load predicted value data storage unit 53, and transmits it to the input / output terminal 60. . In the present embodiment, the output unit 513 generates display screen data indicating the total memory usage, the application memory usage, and the predicted usage according to the elapsed time for each server identifier as the predicted value result display data. Further, the display screen data includes a simulation setting button and an end button. Then, the input / output terminal 60 that has received the predicted value result display data displays a display screen on the display. This display screen displays data on the total memory usage, application memory usage, and predicted usage in the first to third servers. Thus, the load prediction process is completed.

(Simulation process)
Next, a simulation process for optimizing the servers constituting the migration destination system 30 will be described using the predicted usage calculated in the load prediction process described above. This simulation process includes three steps of a simulation condition acquisition process, a simulation calculation process, and a simulation result output process.

  This simulation process is performed after the above-described load prediction process. Specifically, when the simulation setting button on the display screen displayed on the display is selected in the output process of the predicted load value, the simulation process is executed.

  When the predicted usage amount displayed on the display screen exceeds the physical memory amount of each server in the normal operation time period (for example, 9:00 to 11:00 and 15:00 to 17:00), the administrator It is determined that it is necessary to add a server for the system 30, and a simulation setting button is selected. In the present embodiment, it is assumed that the predicted usage amount of the first server does not exceed the physical memory amount, and the predicted usage amounts of the second server and the third server exceed the physical memory amount.

  The input / output terminal 60 that has detected the selection of the simulation setting button notifies the control unit 51 of the load prediction server 50 to that effect. The controller 51 executes simulation condition acquisition processing. Specifically, the simulation condition acquisition unit 514 of the control unit 51 acquires a simulation condition setting screen from the load prediction server 50 and displays it on the display. This simulation condition setting screen is a screen for setting a condition for distributing the predicted load value exceeding the physical memory amount to other servers. This simulation condition setting screen includes a setting condition item field and a setting button. In the setting condition item column, the server identifier of the distribution source server that bears part of the predicted usage amount elsewhere, the ratio of the predicted usage amount to be borne (burden allocation rate), the number of additional servers, the distribution source other than the additional server Information for specifying a distribution destination server that bears the predicted usage amount of the server and item fields for setting conditions regarding load distribution in the distribution destination server are included.

Here, the administrator sets simulation conditions on the simulation condition setting screen. For example, the simulation conditions are input to the simulation condition setting screen as follows.
Enter the server identifiers of the second and third servers as the server identifier of the distribution source server. -Enter 50% of the predicted usage of each distribution source server as the burden allocation rate.
・ Enter “1” as the number of additional servers.
-Enter the first server as the distribution destination server other than the expansion server.
-As an allocation in the distribution destination server, an expansion server: first server = 4: 1 is input.

After the conditions are input in this way, when the setting button on the simulation condition setting screen is selected, the input / output terminal 60 transmits the conditions set on the simulation condition setting screen to the load prediction server 50. The control unit 51 of the load prediction server 50 temporarily records the received simulation conditions in the memory. This simulation condition includes the server identifier of the distribution source server, the predicted usage to be borne, the number of additional servers, the server identifier of the distribution server, and the distribution. Then, the simulation condition acquisition unit 514 assigns a server identifier to the set additional server and temporarily records it in the memory.

  Next, the control unit 51 executes a simulation calculation process. Here, the control unit 51 extracts the load predicted value data from the load predicted value data storage unit 53, and calculates the simulation result predicted use amount in each server according to the recorded simulation conditions. Here, the control unit 51 sequentially extracts the predicted load value data and executes the following series of processes.

  First, the simulation execution means 515 of the control unit 51 extracts the load predicted value data of the distribution source server (second or third server) at the same time from the load predicted value data storage unit 53. The simulation execution means 515 extracts the application memory usage at the same time from the usage history data storage unit 52. Then, the simulation execution means 515 calculates the load distributable amount of the distribution source server by substituting the predicted use amount of the extracted predicted load value data 530 and the application memory use amount into the load distributable amount calculation formula. , Temporarily record in memory. Further, the simulation executing means 515 calculates the predicted usage amount to be borne by multiplying the predicted load increase value by the burden allocation rate.

  Next, the simulation execution means 515 calculates the predicted usage amount of the distribution source server (second or third server). Specifically, the simulation execution means 515 calculates a value obtained by subtracting the calculated predicted usage to be borne from the load predicted value data of the distribution source server at this time as the simulation result predicted usage of the distribution source server. . The simulation execution unit 515 records the calculated simulation result predicted usage amount in the simulation result data storage unit 54 together with the server identifier of each distribution source server and its time.

  Next, the simulation execution means 515 calculates the predicted usage amount of the distribution destination server (first server and additional server). Here, first, the simulation execution unit 515 calculates the total burden amount for each time by summing the burden distributable amounts of the respective distribution source servers (here, the second server and the third server) for each time. Next, the simulation execution unit 515 calculates the burden amount of each distribution destination server (additional server and first server) according to the set distribution. Here, first, the simulation execution means 515 calculates the distribution rate of each distribution destination server according to the distribution set on the simulation condition setting screen. Specifically, the simulation execution means 515 determines the distribution rate “4 / (4 + 1)” for the additional server and the distribution rate “1 / (4 + 1) for the first server according to the set distribution (4: 1). ) ”Is calculated. Then, the simulation execution means 515 calculates the burden amount for each time by multiplying the total burden amount by the distribution rate.

  Then, the simulation execution unit 515 calculates the simulation result predicted usage amount of the distribution destination server using the calculated burden amount. Specifically, the simulation execution unit 515 acquires the predicted usage amount at this time in the distribution destination server from the load predicted value data storage unit 53. Here, when the distribution destination server is an expansion server, the simulation execution means 515 uses the fixed usage amount as the predicted usage amount of the distribution destination server. In the present embodiment, for example, the fixed usage amount of the second server (application memory amount + OS memory usage amount) is used as the fixed usage amount. The simulation execution means 515 adds the calculated burden amount to the predicted usage amount of the distribution destination server, and calculates the simulation result predicted usage amount of the distribution destination server. The simulation execution unit 515 records the calculated simulation result predicted usage amount in the simulation result data storage unit 54 together with the server identifier of each distribution destination server and its time.

And the control part 51 performs a simulation result output process. Specifically, the output unit 513 of the control unit 51 acquires the simulation result predicted usage amount for each server identifier recorded in the simulation result data storage unit 54. Then, the output unit 513 generates simulation predicted value result display data including the acquired simulation result predicted usage amount and transmits it to the input / output terminal 60. In the present embodiment, the output means 513 displays, as the predicted value result display data, display screen data indicating the total memory usage, the application memory usage, the predicted usage, and the simulation result predicted value according to the elapsed time for each server identifier. Generate. Further, the display screen data includes a simulation setting button and an end button. Then, the input / output terminal 60 that has received the predicted value result display data displays a display screen on the display.

  At this time, on the display screen displayed on the input / output terminal 60, for example, as shown in FIG. 5, a table showing the simulation result predicted usage of the first to third servers and the additional server is displayed. In FIG. 5, (a) to (c) respectively indicate usage amounts of the first to third servers. The thin lines in (a) to (c) are the total memory usage a1, b1, c1 and the application memory usage a2, b2, c2 on the peak day, and the middle line is the predicted usage of the migration destination system 30 after the migration. The thick lines are the simulation result predicted usage amounts a4, b4, and c4. Furthermore, the thick line in (d) of FIG. 5 is the simulation result prediction use amount d4 of the additional server.

By viewing this display screen, the administrator can increase the number of servers and grasp the predicted usage as a result of simulation with the set distribution.
According to this embodiment, the following effects can be obtained.

  In the present embodiment, the control unit 51 of the load prediction server 50 executes basic information acquisition processing (step S1-1), executes peak time identification processing (step S1-3), and calculates a scale factor. Processing is executed (step S1-4). In this scale factor calculation process, the control unit 51 calculates the scale factor using the number of login users of the migration source system 20 at the peak time and the number of login users of the migration destination system 30 at the same time. Next, the control part 51 performs the calculation process of a load predicted value (step S1-5). In the calculation process of the predicted load value, the control unit 51 calculates a fixed usage amount, calculates a predicted usage amount using the fixed usage amount and the entire memory usage amount of the extracted usage amount data, and loads the load. The predicted value data 530 is recorded in the load predicted value data storage unit 53. Next, the control part 51 performs the output process of a load predicted value (step S1-6). Therefore, the migration when the users of the migration source system 20 are migrated to the migration destination system 30 using the number of users of the migration source system 20 and the migration destination system 30 and the usage amount actual data 320 of the migration destination system 30. The memory usage of the later migration destination system 30 can be predicted. Therefore, even if the memory usage in the migration source system 20 is unknown, if the number of users is known, the memory usage in the migration destination system 30 after migration can be predicted.

  In the present embodiment, in the load predicted value calculation process (step S1-5), the control unit 51 calculates the total usage amount of the extracted usage amount actual data, the calculated fixed usage amount, and the scale factor as (2 Substitute into the formula to calculate the predicted usage. For this reason, since the predicted usage is calculated by multiplying the value obtained by subtracting the fixed usage that does not depend on the number of users from the total memory usage, the total usage amount of the migration destination system is predicted more accurately. be able to.

-In this embodiment, the control part 51 specifies a peak day in the acquisition process (step S1-2) of whole memory usage, and the peak time in a peak day is specified in the peak time specification process (step S1-3). Identify. The control unit 51 calculates the scale factor using the number of login users of the migration source system 20 at the peak time and the number of login users of the migration destination system 30 at the same time (step S1-4). Therefore, the predicted usage amount of the migration destination system 30 after the migration is specified by using the usage amount actual data 320 on the peak day when the number of users using the migration source system 20 and the migration destination system 30 is the largest. Further, on the peak day, the scale factor is calculated using the number of login users at the time when the sum of the number of login users of the migration source system 20 and the number of login users of the migration destination system 30 is the largest. Therefore, a predicted value close to the maximum usage amount of the migration destination system after the migration can be calculated by using the actual usage amount when the usage amount becomes the largest. Furthermore, by calculating the scale factor using the number of users when the ratio of the number of one login to the whole is as small as possible, a more accurate usage amount can be calculated.

  In the present embodiment, when the input / output terminal 60 detects selection of the simulation setting button, the control unit 51 of the load prediction server 50 executes simulation processing. In this simulation process, the control unit 51 executes a simulation calculation process for calculating and recording a simulation result predicted usage amount obtained by distributing the predicted usage amount of each server using the simulation conditions acquired in the simulation condition acquisition processing. And the control part 51 performs the simulation result output process which produces | generates the simulation prediction value result display data containing a simulation result prediction usage amount, and transmits to the input / output terminal 60. FIG. Thereby, the administrator can grasp the result of the simulation in which the predicted usage amount is distributed by the set expansion server and distribution.

Moreover, you may change the said embodiment as follows.
In the above embodiment, the control unit 51 of the load prediction server 50 calculates the scale factor at the peak time of the peak day, and calculates the predicted usage amount using this scale factor. The time for calculating the scale factor is not limited to this, and may be determined by other methods. For example, the scale factor may be calculated using the number of users at the time when the number of login users of the migration destination system 30 is the largest.

  In the above embodiment, the control unit 51 of the load prediction server 50 calculates the predicted usage at each time using the scale factor at the peak time on the peak day. The method of calculating the scale factor is not limited to this. The scale factor is calculated by calculating the scale factor for each time and using the total memory usage and the scale factor at the same time to calculate the predicted usage. Also good.

In the above embodiment, the control unit 51 of the load prediction server 50 counts the number of login users using the log data of the log data storage units 21 and 31, and logs in users of the migration source system 20 and the migration destination system 30 Got the number. The method for acquiring the number of logged-in users is not limited to this, and the control unit 51 acquires, from the input / output terminal 60, data related to the peak number of the migration source system and the number of users of the old system used for calculating the scale factor. Also good. Specifically, the basic information acquisition unit 511 of the control unit 51 includes a load prediction that includes an input field for inputting the number of users on the peak day and the peak time zone with the largest number of users to use as information specifying the migration source system. The information acquisition screen data is transmitted to the input / output terminal 60, and the load prediction information acquisition screen is displayed. On this load prediction information acquisition screen, the administrator inputs the number of users on the peak day of the migration source system and the peak time zone, and selects the load prediction button. Thereby, the control unit 51 acquires the number of logged-in users in the log data storage unit 31 of the transfer destination system 30 in the peak time zone. The control unit 51 calculates the scale factor by substituting the number of users input on the load prediction information acquisition screen and the number of logged-in users in the log data storage unit 31 into the equation (1). In this case, even if the migration source system does not actually record log data, the predicted usage can be calculated using the number of users. Further, the control unit 51 may store a user number distribution program. This user number distribution program calculates the number of login users for each time by distributing the number of login users around the peak time period from the acquired number of users and the peak time period. When acquiring the number of login users and the peak time zone via the load prediction information acquisition screen, the control unit 51 uses the user number distribution program to distribute the number of users and calculates the number of login users for each time. Then, the control unit 51 may calculate the peak time from the sum of the distributed number of users and the number of login users of the migration destination system 30, and may calculate the scale factor using the number of login users at the peak time. .

  In the simulation process of the above-described embodiment, a simulation condition acquisition process for distributing the predicted usage calculated in the load prediction process is performed, the simulation calculation process is performed, and the simulation result output process for outputting the predicted simulation usage is performed. Was executed. Instead of this simulation processing, the number of additional servers and the load allocation rate that the predicted usage amount does not exceed the physical memory amount of each server may be calculated. In this case, the control unit 51 executes a simulation execution possibility determination process instead of the load prediction value calculation process (step S1-6) in the load prediction process. When the simulation execution is determined, the control unit 51 may automatically perform the simulation following the load prediction process.

  For example, the distribution destination server that distributes the predicted usage amount of the server that exceeds the physical memory amount is only the expansion server, and the predicted usage amount to be borne can be calculated on the assumption that the expansion server is equally allocated. Specifically, the simulation execution means 515 of the control unit 51 stores data relating to the physical memory amount (limit usage amount) of the memory provided in the server and the normal operation time zone. In the following description, it is assumed that the migration destination system is composed of a plurality of servers.

  After executing the load predicted value calculation process (step S1-5), the control unit 51 performs the simulation execution possibility determination process (step S2) shown in FIG. 6 instead of the load predicted value calculation process (step S1-6). -1) is executed.

  In this simulation execution possibility determination process (step S2-1), the control unit 51 determines whether or not the predicted use amount calculated in the load prediction process exceeds the physical memory amount in the normal operation time zone. Specifically, the simulation execution means 515 of the control unit 51 compares the predicted usage amount of the load predicted value data 530 and the physical memory amount in the normal operation time zone.

  Here, if there is no predicted usage exceeding the physical memory amount in the normal operation time zone (“NO” in step S2-1), the simulation execution means 515 indicates that no additional server is required. An output process for transmitting display screen data including a message to the input / output terminal 60 is executed. The input / output terminal 60 displays a display screen based on the display screen data.

  On the other hand, when there is a predicted usage amount that exceeds the physical memory amount in the normal operating time zone (in the case of “YES” in step S2-1), the control unit 51 executes the specifying process of the distribution source server (step S12-1). S2-2). Specifically, the simulation execution means 515 of the control unit 51 temporarily stores a server associated with the predicted usage amount exceeding the physical memory amount in the memory as a distribution source server.

Next, the simulation execution means 515 executes a process for calculating the number of additional servers (step S2-3). Specifically, the simulation execution unit 515 calculates the individual excess usage amount obtained by subtracting the physical memory amount from the maximum value of the predicted usage amount in the normal operation time period in each distribution source server. When there is one distribution source server, this individual excess usage is used as the excess usage. When there are a plurality of distribution source servers, the simulation execution unit 515 calculates a value obtained by summing the individual excess usage amounts of the distribution source servers at the same time, and uses this maximum value as the excess usage amount. On the other hand, the simulation execution means 515 calculates the distribution underwriting capacity obtained by subtracting the fixed usage amount from the physical memory amount in the distribution destination server. Then, the simulation execution means 515 divides the calculated excess usage amount by the distribution undertaken amount, and determines a value obtained by rounding up the value after the decimal point as the number of distribution destination servers (number of expansion servers) to be added.

  Next, the simulation execution means 515 executes a load allocation rate calculation process for each distribution source server (step S2-4). Specifically, the simulation execution means 515 calculates a value obtained by dividing the maximum value of the individual excess usage amount of each distribution source server by the physical memory amount as a burden allocation rate, and temporarily stores it in the memory. Thereby, the simulation execution means 515 can acquire the distribution source server identifier, the burden allocation rate, the number of expansion servers, and the distribution to the expansion servers (distribution destination servers).

  And the control part 51 performs a simulation calculation process similarly to the said embodiment (step S2-5). Further, similarly to the simulation result output process of the above embodiment, the control unit 51 records the calculated simulation result predicted usage and executes the simulation result output process (step S2-6). In this simulation result output process, the control unit 51 generates simulation predicted value result display data including information on the calculated number and distribution of additional servers together with the acquired simulation result predicted usage amount, and transmits it to the input / output terminal 60. . According to this, when the predicted usage exceeds the physical memory amount in the normal operation time zone, the control unit 51 automatically calculates the number of additional servers and the load allocation ratio that can be accommodated in the physical memory amount of the memory. And output. Therefore, the administrator can easily grasp the required additional server and its load allocation ratio when the predicted usage amount of the migration destination system 30 after migration is large.

  In the embodiment described above, the control unit 51 of the load prediction server 50 has predicted the memory usage. For example, the usage amount of another processing apparatus such as a usage amount of a CPU may be calculated.

The schematic of the load prediction system in an embodiment. It is explanatory drawing explaining the structure of the data recorded on the data storage part, (a) is a usage-amount actual data storage part, (b) is a load predicted value data storage part. Explanatory drawing for demonstrating the peak time which calculates a scale factor in a peak day. The flowchart for demonstrating the process sequence of a load prediction process. It is explanatory drawing explaining the display screen displayed by a simulation result output process, (a)-(c) is the total memory usage of 1st-3rd server, (d) is the total memory usage of an expansion server. Show. The flowchart for demonstrating the process sequence of the simulation process in the example of a change.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 ... Client terminal, 20 ... Migration source system, 21 ... Log data storage part, 30 ... Migration destination system, 31 ... Log data storage part, 32 ... Usage amount data storage part, 50 ... Load prediction server, 51 ... Control part 52 ... Usage amount data storage unit, 53 ... Load prediction value data storage unit, 54 ... Data storage unit, 60 ... I / O terminal, 320 ... Usage amount result data, 511 ... Basic information acquisition means, 512 ... Load prediction value Calculation means, 513 ... output means, 514 simulation condition acquisition means, 515 ... simulation execution means, 530 ... load predicted value data.

Claims (6)

Migration source login number data storage means storing the login date and logout date and time of the migration source user using the migration source system;
Migration destination login number data storage means storing the login date and logout date and time of the migration destination user using the migration destination system used in combination with the migration source system ;
Usage amount data storage means for recording the usage amount actually used in the server constituting the migration destination system together with the time;
The state in which all the users using the migration source system are migrated to use the migration destination system from the state where the migration source system and the migration destination system are used together. A usage amount prediction system comprising a control means for predicting the usage amount of the destination system,
The control means includes
From the migration source logins data storage means to identify the number of the source user at a particular time and date, the number of target users in the specified date and time specified from the destination log number of data storage means, the source number of users who the specific , A scale factor calculating means for calculating a scale factor obtained by dividing by the specified number of migration destination users,
A prediction unit that obtains the usage amount actual data from the usage amount actual data storage unit and calculates a usage amount of the migration destination system using a value obtained by multiplying the usage amount actual result and the calculated scale factor; A usage prediction system characterized by that.   The predicting means calculates an increased load amount by multiplying a value obtained by subtracting a fixed use amount from a use amount result obtained from the use amount result data storage means, and the scale factor, and uses the increased load amount as the use amount. The usage amount prediction system according to claim 1, wherein the usage amount is calculated by adding an actual amount. The control means includes
The number of migration source users is obtained from the migration source login number data storage means, the number of migration destination users is obtained from the migration destination login number data storage means, and the number of migration source users and the number of migration destination users are determined daily. To identify the peak day with the highest number of logins,
A peak time specifying means for calculating a total of the number of migration source users and the number of migration destination users on the peak date, and specifying a time with the largest total number of users as a peak time;
The scale factor calculation means acquires the number of migration source users and the number of migration destination users at the specified peak time, calculates a scale factor using the number of migration source users and the number of migration destination users,
The usage amount prediction system according to claim 1, wherein the prediction unit calculates the usage amount using a usage amount record on the peak day. The migration destination system includes a plurality of servers,
The control means stores a normal operation time zone and a limit usage amount of the server,
In the usage amount data storage means, data relating to the usage amount is recorded for each server,
The prediction unit obtains the usage amount results for each server from the usage amount result data storage unit, calculates the usage amount for each server using these usage amount results,
The control means includes
Means for comparing the predicted usage during normal operating hours with the limit usage of the server;
If the predicted usage is greater than or equal to the limit usage, a means for identifying a server associated with the predicted usage as another distribution source server;
If there is one distribution source server, calculate the excess usage amount by subtracting the limit usage amount from the maximum predicted usage amount of the distribution source server.
If there are multiple distribution source servers, subtract the limit usage from the maximum predicted usage of each distribution source server, add them up at each time, and use this total maximum predicted usage as excess usage. Means to identify;
Means for calculating a distribution undertaken amount obtained by subtracting a fixed use amount from the limit use amount;
Means for dividing the calculated excess usage amount by the amount that can be distributed and determining and recording the number of additional servers;
Means for dividing a value obtained by subtracting the limit usage from the maximum value of the predicted usage of each distribution source server by the limit usage, and calculating a burden allocation rate in the distribution source server;
The usage amount prediction system according to any one of claims 1 to 3, further comprising means for outputting the recorded number of additional servers and the burden allocation rate. Migration source login number data storage means storing the login date and logout date and time of the migration source user using the migration source system;
Migration destination login number data storage means storing the login date and logout date and time of the migration destination user using the migration destination system used in combination with the migration source system ;
Usage amount data storage means for recording the usage amount actually used in the server constituting the migration destination system together with the time;
The state in which all the users using the migration source system are migrated to use the migration destination system from the state where the migration source system and the migration destination system are used together. A method for predicting usage using a usage prediction system comprising a control means for predicting usage of a migration destination system,
The control means includes
From the migration source logins data storage means to identify the number of the source user at a particular time and date, the number destination user in said specific time is specified from the target number of login data storage means, the source number of users who the specific , A scale factor calculation stage for calculating a scale factor obtained by dividing by the specified number of migration destination users,
Obtaining the actual usage amount from the actual usage amount data storage unit, and using the value obtained by multiplying the actual usage amount by the calculated scale factor, performing a prediction step of calculating the usage amount of the destination system A usage amount prediction method characterized by: Migration source login number data storage means storing the login date and logout date and time of the migration source user using the migration source system;
Migration destination login number data storage means storing the login date and logout date and time of the migration destination user using the migration destination system used in combination with the migration source system ;
Usage amount data storage means for recording the usage amount actually used in the server constituting the migration destination system together with the time;
The state in which all the users using the migration source system are migrated to use the migration destination system from the state where the migration source system and the migration destination system are used together. A program for predicting the usage amount using a usage amount prediction system comprising a control means for predicting the usage amount of the migration destination system,
The control means;
From the migration source logins data storage means to identify the number of the source user at a particular time and date, the number of target users in the specified date and time specified from the destination log number of data storage means, the source number of users who the specific , A scale factor calculation unit that calculates a scale factor divided by the specified number of migration destination users, and obtains the usage amount result from the usage amount result data storage unit, and the usage amount result and the calculated scale factor A usage amount prediction program that functions as a prediction unit that calculates the usage amount of the migration destination system by using a value obtained by multiplying.

Images