JP2011141908A - System usage rate managing device and system usage rate managing method and program used for the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a system usage rate managing device which easily enables managing system performance and optimizing the number of operation nodes for a computer system constructed by a plurality of computation nodes with different performances and architectures. <P>SOLUTION: A means 111 for setting the maximum performance of the node of a processing system 11 stores the maximum performance of each node constructing an external system into a maximum node performance storage 121 of a storage device 12. A means 112 for acquiring resource usage rate acquires the resource usage rate 101 of each node and imparts it to a virtual usage rate calculating section 113. The virtual usage rate calculating section 113 calculates the virtual usage rate of the system from the maximum performance of each node stored in the maximum node performance storage 121 and the imparted usage rate of the resource of each node, and imparts the virtual usage rate to an output section 114. The output section 114 outputs the imparted virtual usage rate to an output device 13. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a system usage rate management apparatus, a system usage rate management method used therefor, and a program thereof, and more particularly to a management method for managing the performance and usage rate of a node group for a system composed of a plurality of computing nodes.

  Conventionally, as shown in FIG. 9, the system usage rate management apparatus includes a processing device 51 and a storage device 52. The processing device 51 uses a CPU (central processing unit) usage rate acquisition means 511 and a CPU. The rate comparison unit 512 and the HW (hardware) configuration change determination unit 513 are provided, and the storage device 52 includes a CPU usage rate storage unit 521 and a service level contract storage unit 522 (see, for example, Patent Document 1). .

  The conventional system usage rate management apparatus 5 having the above-described configuration operates as follows. First, the CPU usage rate acquisition unit 511 acquires the CPU usage rate of each node from an external system (not shown) including one or more nodes (not shown), and stores it in the CPU usage rate storage unit 521. The service level contract storage unit 522 stores in advance the upper limit of the CPU usage rate to be maintained by each node, the maximum number of nodes that can be increased when the CPU usage rate of each node exceeds this upper limit, and the like.

  The CPU usage rate comparison unit 512 compares the CPU usage rate of each node held by the CPU usage rate storage unit 521 with the upper limit of the CPU usage rate to be maintained held by the service level contract storage unit 522, and the comparison result is HW. This is given to the configuration change determination means 513. When there is a node in which the usage rate exceeding the CPU usage rate to be maintained exists, the HW configuration change determination unit 513 determines that the service level is not satisfied, and augments the node with respect to the external system. At this time, the number of nodes can be increased up to the maximum number of nodes that can be increased which the service level contract storage unit 522 holds.

  On the other hand, when there is only a node having a CPU usage rate or less to be maintained, the HW configuration change determination unit 513 determines that the external system satisfies the service level. If the service level can be maintained even if the number of nodes is reduced, the number of nodes may be reduced. At this time, the HW configuration change determination unit 513 assumes that the calculation capability of each node is the same, and estimates the usage rate after the enhancement or reduction assuming that the product of the usage rate and the number of nodes is constant. For example, when all four nodes are operated at a CPU usage rate of 25%, the estimated usage rate of each node when the number of nodes is increased to 5 is 20%. Conversely, when the number of nodes is reduced to 2, the estimated usage rate of each node is 50%.

JP 2002-024192 A JP-A-10-187636 JP 2002-268922 A

  The above-described conventional system usage rate management apparatus has a problem that when there are nodes having different performances, it cannot be appropriately determined whether or not the HW should be increased. This means that if the usage rate of a low-performance node is higher than the allowable value and the usage rate of a high-performance node is low, the system as a whole will try to reinforce the node even though there is a large capacity Because.

  At this time, a simple performance value such as the number of CPU clocks often does not accurately represent the performance of an application operating on the system. This is because the CPU and computer architectures are diversified, and even with the same number of clocks, the performance varies greatly depending on the architecture and the characteristics of the application to be processed.

  Therefore, the object of the present invention is to solve the above problems and easily manage system performance and optimize the number of operating nodes for a computer system composed of a plurality of computing nodes having different performance and architecture. The present invention provides a system usage rate management apparatus, a system usage rate management method used therefor, and a program therefor.

The system usage rate management apparatus according to the present invention is a system usage rate management that calculates the CPU usage rate of a virtual node when a plurality of computation nodes that perform the same or related tasks and are grouped are regarded as one virtual node. A device,
Node maximum performance setting means for setting the maximum number of requests that each of the plurality of computing nodes can process per unit time as maximum performance; resource usage rate acquiring means for acquiring the CPU usage rate of each of the plurality of computing nodes; A virtual usage rate that calculates the weighted average of the CPU usage rate acquired by the resource usage rate acquisition unit as the CPU usage rate of the virtual node by regarding the maximum performance set by the maximum performance setting unit as a weight. And a calculating means.

The system usage rate management method according to the present invention is a system usage rate management for calculating the CPU usage rate of a virtual node when a plurality of calculation nodes that perform the same or related business and are grouped are regarded as one virtual node. A system usage rate management method used in an apparatus,
A first step of setting, as a maximum performance, a maximum number of requests that each of the plurality of computing nodes can process per unit time; and a second step of acquiring a CPU usage rate of each of the plurality of computing nodes. And calculating the weighted average of the CPU usage rate acquired in the second step as the CPU usage rate of the virtual node by regarding the maximum performance set in the first step as a weight. The third step is executed.

A program according to the present invention is a computer in a system usage rate management apparatus that calculates a CPU usage rate of a virtual node when a plurality of computational nodes that perform the same or related tasks and are grouped are regarded as one virtual node. A program to be executed,
A first process for setting a maximum number of requests that each of the plurality of calculation nodes can process per unit time as a maximum performance; a second process for acquiring a CPU usage rate of each of the plurality of calculation nodes; And a third process of calculating the weighted average of the CPU usage rate acquired in the second process as the CPU usage rate of the virtual node by regarding the maximum performance set in the process as a weight. It is characterized by.

  In other words, the system usage rate management apparatus of the present invention calculates the performance and usage rate of a group of nodes for a computer system composed of a plurality of computing nodes having different performance and architecture, thereby managing and operating the system performance. The number of nodes can be easily optimized.

  More specifically, in the system usage rate management apparatus of the present invention, a node maximum performance setting unit that sets the maximum performance of each node, a resource usage rate acquisition unit that acquires the resource usage rate of each node, and each node Virtual usage rate calculating means for calculating the virtual performance and usage rate of the node group from the maximum performance and resource usage rate, and output means for outputting the calculated overall performance and usage rate. The above purpose is achieved by outputting or using the usage rate.

  As a result, the system usage rate management device of the present invention can calculate the performance and usage rate of a node group for a computer system composed of a plurality of nodes having different performances and architectures. It becomes possible to quickly grasp the load.

  Further, in the system usage rate management apparatus of the present invention, the allowable usage rate of the node group can be calculated from the allowable usage rate of each node, so it is possible to quickly check whether the system can be operated with the usage rate within the allowable range as a whole. It becomes possible to judge.

  Furthermore, since the system usage rate management apparatus according to the present invention can output the virtual usage rate and the time at which the virtual usage rate is calculated, it is possible to create an operation report that summarizes the virtual usage rate in time series.

  With the configuration and operation as described above, the present invention easily manages system performance and optimizes the number of operating nodes for a computer system composed of a plurality of computing nodes having different performance and architecture. The effect that it can be obtained.

It is a block diagram which shows the structure of the system usage rate management apparatus by this invention. It is a flowchart which shows operation | movement of the system utilization rate management apparatus by this invention. It is a block diagram which shows the structure of the system usage rate management apparatus by the 1st Embodiment of this invention. It is a flowchart which shows operation | movement of the system usage rate management apparatus by the 1st Embodiment of this invention. It is a block diagram which shows the structure of the system usage rate management apparatus by the 2nd Embodiment of this invention. It is a flowchart which shows operation | movement of the system usage rate management apparatus by the 2nd Embodiment of this invention. It is a block diagram which shows the structure of the system usage-rate management apparatus by the 3rd Embodiment of this invention. It is a flowchart which shows operation | movement of the system usage rate management apparatus by the 3rd Embodiment of this invention. It is a block diagram which shows the structure of the system usage rate management apparatus relevant to this invention.

  Next, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a system usage rate management apparatus according to the present invention. In FIG. 1, the system usage rate management device 1 includes a processing device 11, a storage device 12, and an output device 13. The processing device 11 includes a node maximum performance setting unit 111, a resource usage rate acquisition unit 112, a virtual usage rate calculation unit 113, and an output unit 114, and the storage device 12 includes a node maximum performance storage unit 121. 11 stores a program executed by the computer 11 (program executable by a computer).

  The node maximum performance setting unit 111 stores the maximum performance of each node (not shown) constituting the external system in the node maximum performance storage unit 121. The resource usage rate acquisition unit 112 acquires the resource usage rate 101 of each node and gives it to the virtual usage rate calculation unit 113. The virtual usage rate calculation unit 113 calculates the virtual usage rate of the system from the maximum performance of each node held in the node maximum performance storage unit 121 and the resource usage rate of each assigned node, and the virtual usage rate The rate is given to the output unit 114. The output unit 114 outputs the assigned virtual usage rate to the output device 13.

  FIG. 2 is a flowchart showing the operation of the system usage rate management apparatus 1 according to the present invention. The operation of the system usage rate management apparatus 1 according to the present invention will be described with reference to FIG. 1 and FIG. Note that the processing shown in FIG. 2 is realized by the processing device 11 executing a program in the storage device 12.

  First, the node maximum performance setting unit 111 stores the maximum performance of each node constituting the external system in the node maximum performance storage unit 121 (step S1 in FIG. 2). Here, the maximum performance refers to the number of requests that can be processed per unit time. Hereinafter, in addition to requests to the Web server and database server, requests that collectively generate processing on the receiving side, such as packets sent and received via the network, machine language, statements in program text, functions, etc. Call it.

  As the maximum performance of each node, a value input from a user through an input device (not shown), a value prepared in advance, a value calculated independently, or the like is used. Further, for reasons described later, the maximum performance of each node is not an absolute value, and may be a ratio of the maximum performance between nodes.

  In the external system, nodes that perform the same type or related work may be grouped, and the node maximum performance setting or the virtual usage rate may be calculated for the nodes in the same group. Hereinafter, unless otherwise specified, the entire node group in the same group and the entire system will be described without distinction.

  The resource usage rate acquisition unit 112 acquires the resource usage rate 101 of each node and gives it to the virtual usage rate calculation unit 113 (step S2 in FIG. 2). The resource usage rate includes a CPU usage rate and a disk usage rate. The usage rate refers to a rate of time during which a resource such as a CPU (central processing unit) or a disk is actually used per unit time.

  The virtual usage rate calculation unit 113 calculates the virtual usage rate of the node group from the maximum performance of each node held in the node maximum performance storage unit 121 and the resource usage rate of each assigned node. A typical usage rate is given to the output unit 114 (step S3 in FIG. 2).

  When the maximum performance of the node having the identification number i is αi · μ, the maximum performance of the node group (node group) can be expressed by the sum Σi (αi · μ) of the maximum performance of each node. Here, αi · μ represents absolute performance, and αi represents relative performance (performance ratio) with other nodes. For example, when the absolute performance of three nodes is 120 requests / second for α1 · μ, 80 requests / second for α2 · μ, and 200 requests / second for α3 · μ, α1 = 3, α2 = 2, α3 = 5.

On the other hand, when the request arrival rate to the node i is λi, the resource usage rate ρi of each node is
ρi = λi / (αi · μ)
It can be expressed as. Here, when the entire node group is considered as one virtual node, the request arrival rate to the node group is:
Σi (λi)
Therefore, the usage rate (virtual usage rate) ρ of the node group is
ρ = Σi (λi) / Σi (αi · μ)
It can be expressed as.

further,
λi = ρi ・ αi ・ μ
By transforming the above equation using
ρ = Σi (ρi · αi · μ) / Σi (αi · μ)
= Σi (ρi · αi) / Σi (αi)
Is obtained. That is, the virtual usage rate is calculated by the above formula using the absolute performance value or the performance ratio between nodes together with the usage rate of each node.

  In addition, if an allowable usage rate is specified for each node that constitutes the calculation target node group from the viewpoint of response time, etc., and the usage rate is below a certain usage rate, the allowable usage rate of each node Is given to the resource usage rate acquisition means 112, and the allowable usage rate of the node group can be obtained.

  Finally, the output unit 114 outputs the virtual usage rate of the node group to the output device 13. At this time, the output unit 114 may output the calculation time, node configuration information, and the like together with the virtual usage rate. The embodiment of the present invention can be operated not only on a computer or a calculation node different from the external system but also on a node included in the external system.

  As described above, according to the present invention, the capacity and usage rate of a node group can be calculated for a computer system composed of a plurality of nodes having different performance and architecture, so that the usage status and load of the system can be quickly grasped. can do.

  In the present invention, since the allowable usage rate of the node group can be calculated from the allowable usage rate of each node, it is possible to quickly determine whether the system can be operated with the usage rate within the allowable range as a whole. .

  Furthermore, in the present invention, since the virtual usage rate can be output together with the virtual usage rate, an operation report in which the virtual usage rates are collected in time series can be created.

  A specific example of the present invention will be described. In the following description, although not shown, it is assumed that the external system includes three nodes on which the Web server operates, two application server nodes, and two database server nodes.

  When calculating the virtual usage rate for three Web servers, the user of the present invention assigns the same group identifier to an identifier such as an IP (Internet Protocol) address or a machine name of the node, thereby grouping the nodes. Can be In addition, using a GUI (Graphical User Interface), the nodes displayed on the screen may be grouped by an operation such as placing them in an area representing the group.

  Regarding the three Web servers grouped in this way, first, when the maximum performance of each node is input from the node maximum performance setting unit 111, it is stored in the node maximum performance storage unit 121. Here, based on requests such as HTTP (Hyper Text Transfer Protocol) and CGI (Common Gateway Interface) that can be processed by the Web server, the maximum performance of the three nodes is “α1 · μ = 120”, “α2 · μ =, respectively. 80 ”and“ α3 · μ = 200 ”.

  The input value may be directly given by the user, or may be given by another device that holds a value that can be used as the maximum performance. The node maximum performance storage unit 121 may hold the maximum performance of each node in the form of a performance ratio. In this case, the performance ratio is “α1 = 3”, “α2 = 2”, and “α3 = 5”.

  The resource usage rate acquisition unit 112 acquires the resource usage rate 101 of each node and gives it to the virtual usage rate calculation unit 113. Here, as an example, a CPU usage rate provided by an OS (Operating System) of an external system is considered as a resource usage rate. Assume that the CPU usage rates acquired from each node are “ρ1 = 0.2”, “ρ2 = 0.15”, and “ρ3 = 0.6”, respectively.

The virtual usage rate calculation unit 113 uses the maximum performance of each node held in the node maximum performance storage unit 121 and the CPU usage rate of each node given from the resource usage rate acquisition unit 112 to virtually use the 3 node web server. The rate is calculated and the virtual usage rate is given to the output unit 114. The virtual usage rate ρ at this time is
ρ = Σi (ρi · αi · μ) / Σi (αi · μ)
= (0.2 · 120 + 0.15 · 80 + 0.6 · 200)
/ (120 + 80 + 200)
= 0.39
It becomes. Even when using the maximum performance ratio,
ρ = Σi (ρi · αi) / Σi (αi)
= (0.2 · 3 + 0.15 · 2 + 0.6 · 5) / (3 + 2 + 5)
= 0.39
Thus, the same result as above is obtained.

  Finally, the output unit 114 outputs the obtained virtual usage rate of the system to the output device 13 such as a display or a printer. At this time, in the present invention, the virtual usage rate of the group can be displayed at a location on the screen corresponding to the grouped nodes on the GUI on which the servers are grouped. In the present invention, the CPU usage rate of each node is calculated, the virtual usage rate is calculated, output, and recorded continuously, so that the change in the virtual usage rate over time is displayed in the form of a graph or the like. be able to.

  Furthermore, in the present invention, a report useful for an administrator or user of an external system is created by collecting the CPU usage rate of each node and the virtual usage rate calculated therefrom in a list or graph. be able to. On the GUI or report, configuration information such as the CPU type, memory capacity, OS, middleware, etc. installed in each node may be output together with the time and virtual usage rate.

  Furthermore, in the present invention, it is possible to create a report that summarizes the transition of the virtual usage rate of each group by performing grouping and calculation of the virtual usage rate in the same manner as described above for the application server and the database server. On the basis of this, it is possible to make an operation plan such as interchange of nodes between groups or increase of nodes.

  FIG. 3 is a block diagram showing the configuration of the system usage rate management apparatus according to the first embodiment of the present invention. In FIG. 3, the system usage rate management apparatus 2 according to the first embodiment of the present invention is provided with a service satisfaction evaluation unit 211 in the processing device 21 and a service level contract storage unit 221 in the storage device 22. Except for the above, the configuration is the same as that of the system usage rate management apparatus 1 according to the embodiment of the present invention shown in FIG. 1, and the same components are denoted by the same reference numerals. In addition, the operation of the same component is the same as that of the embodiment of the present invention.

  The service satisfaction evaluation unit 211 evaluates the virtual usage rate calculated by the virtual usage rate calculation unit 113 based on the service level contract held by the service level contract storage unit 221, and gives the evaluation result to the output unit 114. The storage device 22 stores a program executed by the processing device 21 (a computer executable program).

  FIG. 4 is a flowchart showing the operation of the system usage rate management apparatus 2 according to the first embodiment of the present invention. The operation of the system usage rate management apparatus 2 according to the first embodiment of the present invention will be described with reference to FIGS. Note that the processing shown in FIG. 4 is realized by the processing device 21 executing the program in the storage device 22. Also, the processing operations (steps S11 to S13 shown in FIG. 4 (the operations of the node maximum performance setting unit 111, the resource usage rate acquisition unit 112, and the virtual usage rate calculation unit 113) correspond to the above-described embodiment of the present invention. Since this is the same as the operation of the part (the processing operation of steps S1 to S3 in FIG. 2), the description thereof is omitted.

  In the embodiment of the present invention, the virtual usage rate calculated by the virtual usage rate calculation unit 113 is given to the output unit 114. However, in this embodiment, the virtual usage rate calculated by the virtual usage rate calculation unit 113 is used as a service. It is given to the satisfaction evaluation unit 211.

  The service satisfaction evaluation unit 211 evaluates the virtual usage rate calculated by the virtual usage rate calculation unit 113 based on the service level contract held by the service level contract storage unit 221. The service level contract storage unit 221 holds an allowable usage rate that regulates operation at a certain usage rate or less for a node group. Further, the service level contract storage unit 221 may hold a charging method based on the virtual usage rate of the system, a countermeasure to be taken when the virtual usage rate of the system exceeds the allowable usage rate, and the like. In this case, countermeasures include adding a node and changing a charging method. Further, when the virtual usage rate is significantly lower than the allowable usage rate, the number of nodes may be reduced.

  The service satisfaction evaluation unit 211 determines whether or not the virtual usage rate of the system is equal to or lower than the allowable usage rate (step S14 in FIG. 4), and if the determination result is true, evaluates that the service level is satisfied. As a result, it is given to the output unit 114. When the result of the determination is false, the service satisfaction evaluation unit 211 gives a countermeasure such as a change in the node configuration and billing method held by the service level contract storage unit 221 to the output unit 114 (step S15 in FIG. 4). Finally, the output unit 114 outputs the assigned evaluation result to the output device 13.

  As described above, in this embodiment, the number of nodes can be optimized and the charging method can be reviewed based on the service level contract, so that the work for maintaining and managing the service provided to the user can be simplified. it can.

  A specific example of the first embodiment of the present invention will be described. In the following description, although not shown, it is assumed that the external system includes three nodes on which the Web server operates, two application server nodes, and two database server nodes.

  In the first embodiment of the present invention, the calculated virtual usage rate is given to the service satisfaction evaluation unit 211 instead of the output unit 114. The service satisfaction evaluation unit 211 determines whether or not the virtual usage rate is equal to or lower than the allowable usage rate according to the specified content of the service level contract storage unit 221. In the service level contract storage unit 221, for each service, each node group, and node group, an allowable usage rate and a countermeasure when the allowable usage rate is exceeded are defined. For example, it is assumed that the operation of the Web server node group at a virtual usage rate of 35% or less, the addition of a node when this value is exceeded, and the type of node to be added are defined.

  In a specific example of the present invention, the virtual usage rate of the three nodes for the Web server is 39%, which exceeds the specified value. At this time, if the type of node to be added and the maximum performance are unknown, a predetermined number is added. Conversely, when the maximum performance of the node to be added is known, such as when the node to be added is the same type as a node that has already been deployed, the virtual usage rate after the addition can be estimated.

For example, consider adding one node with a maximum performance of 80. Formula for calculating virtual usage rate,
ρ = Σi (ρi · αi · μ) / Σi (αi · μ)
Since the numerator in FIG. 4 represents the total number of requests Σi (λi), assuming that it does not change after addition, the virtual usage rate after addition is
ρ = Σi (ρi · αi · μ) / Σi (αi · μ)
= (0.2 · 120 + 0.15 · 80 + 0.6 · 200)
/ (120 + 80 + 200 + 80)
= 0.325
Thus, it can be seen that the usage rate is below the allowable usage rate.

Also, let X be the maximum performance of the node to be added,
(0.2 ・ 120 + 0.15 ・ 80 + 0.6 ・ 200)
/(120+80+200+X)≦0.35
Since 45.8 ≦ X is obtained by solving the above, it can be seen that it is sufficient to add one node having a maximum performance of 45.8 or more. Further, nodes may be added so that the total of a plurality of units is 45.8 or more.

  In addition to the node addition instruction, the service satisfaction evaluation unit 211 gives the output unit 114 information such as the expected usage rate after addition obtained as described above, and what level of node should be added. To do. Finally, the output unit 114 outputs these pieces of information to the output device 13 such as a display. If the external system has a mechanism for automatically increasing the number of nodes, an instruction to add a node can be directly given to the external system.

  FIG. 5 is a block diagram showing a configuration of a system usage rate management apparatus according to the second exemplary embodiment of the present invention. In FIG. 5, the system usage rate management device 3 according to the second exemplary embodiment of the present invention includes a request arrival rate acquisition unit 3111, a node maximum performance calculation unit 3112, and a resource usage rate acquisition unit 112. The configuration is the same as that of the system usage rate management apparatus 1 according to the embodiment of the present invention shown in FIG. 1 except that the node maximum performance setting means 311 is provided, and the same components are denoted by the same reference numerals. It is. In addition, the operation of the same component is the same as that of the embodiment of the present invention.

  The request arrival rate acquisition unit 3111 acquires the request arrival rate 201 of each node and gives it to the node maximum performance calculation unit 3112. The node maximum performance calculation unit 3112 calculates the maximum performance of each node from the request arrival rate of each node given from the request arrival rate acquisition unit 3111 and the resource usage rate of each node given from the resource usage rate acquisition unit 112. The maximum performance is stored in the node maximum performance storage unit 121.

  FIG. 6 is a flowchart showing the operation of the system usage rate management apparatus 3 according to the second embodiment of the present invention. The operation of the system usage rate management apparatus 3 according to the second exemplary embodiment of the present invention will be described with reference to FIGS. Note that the processing shown in FIG. 6 is realized by the processing device 31 executing a program in the storage device 12.

  First, the request arrival rate acquisition unit 3111 acquires the request arrival rate 201 of each node, and assigns the request arrival rate 201 to the node maximum performance calculation unit 3112 (step S21 in FIG. 6).

  The resource usage rate acquisition unit 112 acquires the resource usage rate 101 of each node at the same time as or near to the time when the request arrival rate is acquired by the request arrival rate acquisition unit 3111, and calculates the resource usage rate 101 as a virtual usage rate. To the unit 113 and the node maximum performance calculation unit 3112 (step S22 in FIG. 6).

  The execution order of steps S21 and S22 may be any order. Further, the request arrival rate and resource usage rate of each node assigned to the node maximum performance calculation unit 3112 may be an average value calculated from sampling data collected during a certain period.

  The node maximum performance calculation unit 3112 calculates the maximum performance of each node from the assigned request arrival rate and resource usage rate of each node, and stores the maximum performance in the node maximum performance storage unit 121 (step S23 in FIG. 6). .

The maximum performance αi · μ of the node i is equal to the number of requests that can be processed when the usage rate is 1.0 (100%), the request arrival rate to the node i is λi, and the resource usage rate is ρi ,
αi · μ = λi / ρi
Is calculated by The average request arrival rate and the average resource usage rate may be calculated as λi and ρi, respectively.

  The virtual usage rate calculation unit 113 is similar to the operation of the virtual usage rate calculation unit 113 in the embodiment of the present invention described above, and the maximum performance of each node held in the node maximum performance storage unit 121 and each assigned node The virtual usage rate is calculated from the resource usage rate, and the virtual usage rate is given to the output unit 114 (step S24 in FIG. 6).

  Finally, the output unit 114 outputs the assigned virtual usage rate to the output device 13. At this time, the output unit 114 may output the calculated maximum performance αi · μ of each node and the maximum performance Σi (αi · μ) of the node group together.

  The performance of an application operating on the system is not simply proportional to the CPU performance, but strongly depends on the architecture of the CPU and computer, the characteristics of the application, the usage method, and the like. However, in this embodiment, the system performance and virtual usage rate can be calculated based on the maximum performance of each node derived from the operational data, so that more accurate system usage and load can be reduced with less burden. I can grasp it.

  A specific example of the second embodiment of the present invention will be described. First, the request arrival rate acquisition unit 3111 collects the request arrival rate 201 of each node of the external system for a certain period, and calculates the average value of the request arrival rates based on the collected data. The calculated average request arrival rate is given to the node maximum performance calculation unit 3112.

  Similarly, the resource usage rate acquisition unit 112 collects the resource usage rate 101 of each node of the external system for a certain period, calculates the average request arrival rate based on the collected data, and sets the average request arrival rate to the node maximum This is given to the performance calculation unit 3112.

  The node maximum performance calculation unit 3112 calculates the maximum performance of each node from the assigned average request arrival rate and average resource usage rate of each node, and stores the maximum performance in the node maximum performance storage unit 121.

Here, it is assumed that the external node group includes three nodes, the average request arrival rate of each node is “λ1 = 70”, “λ2 = 45”, “λ3 = 320”, and the average resource usage rate is “ρ1 = 0”. .2 ”,“ ρ2 = 0.15 ”, and“ ρ3 = 0.4 ”. At this time, the maximum performance αi · μ of each node is
αi · μ = λi / ρi
From the calculation formula, “α1 · μ = 350”, “α2 · μ = 300”, and “α3 · μ = 800”.

  When the resource usage rate of each node is given from the resource usage rate acquisition unit 112, the virtual usage rate calculation unit 113 refers to the maximum performance value of each node as in the case of the above-described embodiment of the present invention. Calculate the virtual usage rate of the node group. Further, the virtual usage rate can be calculated retroactively with respect to the resource usage rate data of each node acquired at the time of calculating the maximum performance of each node and the average value thereof.

  FIG. 7 is a block diagram showing the configuration of a system usage rate management apparatus according to the third embodiment of the present invention. In FIG. 7, the system usage rate management device 4 according to the third exemplary embodiment of the present invention includes a maximum performance comparison unit 411 and a reliability calculation unit 412 in the processing device 41 instead of the virtual usage rate calculation unit 113. In addition, the system usage rate management according to the second embodiment of the present invention shown in FIG. 5 except that the storage device 42 includes a maximum performance storage unit 421 and a comparison count and effective count storage unit 422. The configuration is the same as that of the device 3, and the same reference numerals are given to the same components. The operation of the same component is the same as that of the second embodiment of the present invention.

  The maximum performance comparison unit 411 uses the maximum node performance value adopted for the calculation of the virtual usage rate held in the adopted maximum performance storage unit 421 and the newly calculated latest value held in the node maximum performance storage unit 121. The maximum node performance value is compared to the current node maximum performance value, and it is checked whether it is valid. The comparison count and the valid count are stored in the comparison count and the valid count. Stored in the unit 422.

  The reliability calculation unit 422 calculates the reliability of the maximum node performance being employed from the comparison count and the effective count stored in the comparison count and effective count storage unit 422, and outputs the reliability of the maximum node performance being employed. To part 114. The storage device 42 stores a program executed by the processing device 41 (a program executable by a computer).

  FIG. 8 is a flowchart showing the operation of the system usage rate management apparatus 4 according to the third embodiment of the present invention. The operation of the system usage rate management apparatus 4 according to the third exemplary embodiment of the present invention will be described with reference to FIGS. Note that the processing shown in FIG. 8 is realized by the processing device 41 executing the program in the storage device 42. Also, the processing operations (operations of the request arrival rate acquisition unit 3111, the resource usage rate acquisition unit 112, and the node maximum performance calculation unit 3112) in steps S31 to S33 in FIG. 8 are the second embodiment of the present invention illustrated in FIG. Since this is the same as the operation of the corresponding part (the processing operation of steps S21 to S23 in FIG. 6), the description thereof is omitted.

  In the present embodiment, the value of the maximum node performance currently employed for calculating the virtual usage rate is held in the maximum performance storage unit 421 being employed. The node maximum performance value to be held is a value calculated in the past, a value given from the outside, or the like. The maximum performance comparison unit 411 compares the node maximum performance being employed with the latest node maximum performance newly calculated this time held in the node maximum performance storage unit 121 (step S34 in FIG. 8).

Specifically, the maximum performance comparison unit 411 calculates the node group performance (virtual maximum performance) calculated as the total value of the adopted node maximum performance values and the virtual value calculated from the latest node maximum performance. Compare with the maximum performance value and check whether the difference between the two performance values is within the allowable range. As a simple inspection method, when the virtual maximum performance being adopted is Up, the latest virtual maximum performance is Un, and the allowable error is e, the value of Un is
Up · (1-e) ≦ Un ≦ Up · (1 + e)
If it is within the range, there is a method of determining that the virtual maximum performance being employed is still effective. At this time, a value calculated outside the device according to the present embodiment may be given to the maximum performance comparison unit 411 as the latest node maximum performance.

  The maximum performance comparison unit 411 increases the number of comparisons held in the comparison number and effective number storage unit 422 by 1 (step S35 in FIG. 8). In addition, the process of said step S34 and the process of step S35 may be arbitrary orders.

  As a result of comparing the two virtual maximum performance values, the maximum performance comparison unit 411 increases the effective number held in the comparison number and effective number storage unit 422 by 1 when the virtual maximum performance value being adopted is determined to be valid. (FIG. 8, steps S36 and S37). The maximum performance comparison unit 411 determines that the difference between the virtual maximum performance being adopted and the latest virtual maximum performance is beyond the allowable range, and if the virtual maximum performance being adopted is determined to be invalid, Transition.

The reliability calculation unit 412 calculates the reliability of the maximum node performance being adopted from the comparison number and the effective number held in the comparison number and effective number storage unit 422, and outputs the reliability of the maximum node performance being adopted. To the unit 114 (step S38 in FIG. 8). When the reliability D is Na and the effective number is Nv,
D = Nv / Na
The ratio of the number of effective times to the number of comparisons as calculated in the above is quantified. Finally, the output unit 114 outputs the assigned reliability to the output device 13.

  As described above, in this embodiment, the reliability of the maximum performance value being referred to can be calculated based on the latest maximum performance value. A more accurate virtual usage rate can be calculated by detecting this when the calculated maximum performance value is no longer suitable for the current situation and taking measures such as resetting the maximum performance.

  A specific example of the third embodiment of the present invention will be described. In the present embodiment, the latest node maximum performance stored and calculated in the node maximum performance storage unit 121, and the node maximum performance that has been calculated or input in the past and that is currently used for calculation of the current virtual usage rate, etc. Compare

Regarding the above node group, it is assumed that the maximum node performance being employed is “β1 · μ = 420”, “β2 · μ = 290”, and “β3 · μ = 910”, respectively. At this time, the virtual maximum performance of the node group is
420 + 290 + 910 = 1620
It becomes.

Now, when the latest node maximum performance is calculated as “α1 · μ = 350”, “α2 · μ = 300”, “α3 · μ = 800”, the latest virtual maximum performance is
350 + 300 + 800 = 1450
It becomes. If the allowable error is 10%, the maximum performance comparison unit 411
1620 · (1-0.1) ≦ 1450 ≦ 1620 · (1 + 0.1)
Is checked and the number of comparisons is further increased by one.

in this case,
1620 · (1-0.1) = 1458
Since “1458 ≦ 1450” does not hold, the effective number is not increased. In this way, the node maximum performance calculation unit 3112 and the maximum performance comparison unit 411 repeat the calculation of the latest node maximum performance and the comparison with the node maximum performance being adopted.

It is assumed that the number of comparisons is “382” and the number of effective times is “295” by the calculation and comparison of the maximum node performance this time. At this time, the reliability calculation unit 412 calculates the reliability,
295/382 ≒ 0.772
That is, 77.2% is calculated and given to the output unit 114.

  As a result, if the calculated reliability is very low, it means that the maximum performance of the node being adopted does not match the current load pattern, so it is necessary to take measures such as updating the maximum performance of the node. Become.

  As described above, the performance of an application operating on the system is not simply proportional to the CPU performance, but strongly depends on the CPU and computer architecture, application characteristics, usage, and the like. However, in the present invention, since the performance and usage rate of a node group can be calculated for a computer system composed of a plurality of nodes having different performance and architecture, a more accurate system usage status and load can be quickly determined. I can grasp it.

  In the present invention, since the allowable usage rate of the node group can be calculated from the allowable usage rate of each node, it is possible to quickly determine whether or not the system can be operated with the usage rate within the allowable range as a whole. .

  Furthermore, in the present invention, since the virtual usage rate can be output together with the virtual usage rate, it is possible to create an operation report that summarizes the virtual usage rate in time series.

  In the present invention, the virtual maximum capacity and the virtual usage rate of the entire system can be calculated if all nodes constituting the external system are made one node group. It is also possible to group node groups into higher groups and calculate the virtual maximum capacity and virtual usage rate of the upper group using the virtual maximum capacity and virtual usage rate of each group.

  The present invention can be applied to uses such as performance management, operation management, and capacity planning for parallel computers, cluster systems, and distributed systems composed of a plurality of computing nodes.

1, 2, 3, 4 System usage rate management device 11, 21, 31, 41 Processing device 12, 22, 42 Storage device
13 Output device
101 Node resource utilization
111,311 node maximum performance setting means
112 Resource usage rate acquisition means
113 Virtual usage rate calculation unit
114 Output unit
121 Node maximum performance storage
201 Node request arrival rate
211 Service Satisfaction Evaluation Department
221 Service level contract storage
411 Maximum performance comparator
412 Reliability calculation unit
421 Maximum performance storage in use
422 Comparison number and effective number storage unit
3111 Request arrival rate acquisition means
3112 Node maximum performance calculator

Claims (11)

A system usage rate management apparatus that calculates a CPU usage rate of a virtual node when a plurality of computation nodes that perform the same kind or related tasks and are grouped are regarded as one virtual node,
Node maximum performance setting means for setting the maximum number of requests that each of the plurality of computing nodes can process per unit time as maximum performance; resource usage rate acquiring means for acquiring the CPU usage rate of each of the plurality of computing nodes; A virtual usage rate that calculates the weighted average of the CPU usage rate acquired by the resource usage rate acquisition unit as the CPU usage rate of the virtual node by regarding the maximum performance set by the maximum performance setting unit as a weight. A system usage rate management apparatus comprising: a calculation means.   The node maximum performance setting unit includes the resource usage rate acquisition unit, a request arrival rate acquisition unit that acquires a request arrival rate of each of the plurality of calculation nodes, and the resource usage acquired by the resource usage rate acquisition unit. 2. A node maximum performance calculating unit that calculates and holds a maximum performance of each of the plurality of calculation nodes from the rate and the request arrival rate acquired by the request arrival rate acquiring unit. The system usage rate management device described.   3. A service satisfaction evaluation unit that evaluates whether or not the virtual usage rate calculated by the virtual usage rate calculation unit satisfies a prescribed service level. System usage rate management device.   Maximum performance comparison means for comparing the latest maximum performance set by the node maximum performance setting means with the maximum performance currently employed, and the total number and the latest maximum performance compared by the maximum performance comparison means And reliability calculation means for calculating reliability from the number of times that the error of both values of the maximum performance currently employed is within an allowable range. One of the system usage rate management apparatuses described.   The system usage rate management apparatus according to any one of claims 1 to 4, further comprising means for displaying the calculated virtual usage rate along a time series. A system usage rate management method used in a system usage rate management apparatus for calculating a CPU usage rate of a virtual node when a plurality of computation nodes that perform the same or related tasks and are grouped are regarded as one virtual node. There,
A first step of setting, as a maximum performance, a maximum number of requests that each of the plurality of computing nodes can process per unit time; and a second step of acquiring a CPU usage rate of each of the plurality of computing nodes. And calculating the weighted average of the CPU usage rate acquired in the second step as the CPU usage rate of the virtual node by regarding the maximum performance set in the first step as a weight. A system usage rate management method characterized by executing the third step.   The step of setting the maximum performance on the system usage rate management apparatus side includes the step of acquiring the resource usage rate, the step of acquiring the request arrival rate of each of the plurality of computing nodes, and the acquired resource usage The system usage rate management method according to claim 6, further comprising: calculating and holding a maximum performance of each of the plurality of calculation nodes from the rate and the acquired request arrival rate.   8. The system usage rate management method according to claim 6, further comprising the step of evaluating whether or not the virtual usage rate satisfies a prescribed service level on the system usage rate management apparatus side. .   The step of comparing the latest maximum performance set on the system usage rate management device side with the maximum performance currently employed, the total number of the comparison, the latest maximum performance, and the maximum performance currently employed. The system usage rate management method according to claim 6, further comprising a step of calculating the reliability based on the number of times that the error of both values is within an allowable range.   The system usage rate management method according to any one of claims 6 to 9, wherein the system usage rate management apparatus displays the calculated virtual usage rate along a time series. This is a program to be executed by a computer in the system usage rate management apparatus that calculates the CPU usage rate of a virtual node when a plurality of calculation nodes that perform the same or related tasks and are grouped are regarded as one virtual node. And
A first process for setting a maximum number of requests that each of the plurality of calculation nodes can process per unit time as a maximum performance; a second process for acquiring a CPU usage rate of each of the plurality of calculation nodes; And a third process of calculating the weighted average of the CPU usage rate acquired in the second process as the CPU usage rate of the virtual node by regarding the maximum performance set in the process as a weight. A program characterized by

Images