JP4761229B2 - Operation management apparatus, operation management method and program - Google Patents

Description

  The present invention relates to an operation management apparatus that uses a virtual machine (VM).

  An operation management apparatus capable of controlling a server constituting a system by software by using a virtual machine (VM) has been developed. With conventional operation management devices, it is easy to create and replicate servers, move operating locations, and dynamically change processing performance beyond hardware limitations, so flexible services that follow changes in the environment Expected to be able to deliver.

  However, in the conventional operation management apparatus, the failure is determined only by the performance information collected from the monitoring agent of the server (guest computer) configured with the VM. For this reason, the resource allocation change dynamically performed on the host computer cannot be grasped, and the guest performance abnormality cannot be detected accurately. For this reason, there is a problem that erroneous detection of a failure or omission of detection occurs, and the efficiency of operation management work is greatly reduced.

  Here, we introduce technologies related to operation management.

  Japanese Unexamined Patent Application Publication No. 2006-65430 describes a virtual machine performance changing method (Patent Document 1). The virtual machine performance changing method is characterized in that the CPU performance of the virtual machine is dynamically changed by changing the value of the dummy cycle included in the instruction execution part of the virtual machine according to the change in the load of the virtual machine. .

  Japanese Patent Laid-Open No. 2002-32244 describes a virtual computer (Patent Document 2). The virtual computer includes virtual computer control means for controlling a plurality of guest operating systems having failure information collection means. When the virtual machine control unit becomes unable to control the guest operating system due to its own failure, the virtual machine locks the operating memory area of the guest operating system in which the virtual machine control unit itself is operating, and the memory image of the operating memory area Is further provided with means for aborting itself after saving to the external storage.

  Japanese Patent Laid-Open No. 2005-115751 describes an in-vehicle electronic device (Patent Document 3). The computer system operates on the first OS, a service application that operates on the first OS and performs normal business processing, a second OS that is different from the first OS, and the second OS And an analysis prediction application. The first OS holds the status information and operation record information of the own OS, and the analysis prediction application analyzes the contents of the information held by the first OS and detects a failure sign. .

  Japanese Patent Application Laid-Open No. 2006-39763 describes a guest OS debugging support method (Patent Document 4). The guest OS debugging support method supports debugging of a guest OS that operates in a virtual machine execution environment provided by a virtual machine manager. The guest OS debugging support method includes a step in which a virtual machine manager constructs a second virtual machine execution environment different from the first virtual machine execution environment in which the first guest OS operates, By copying the guest OS to the second virtual machine execution environment, including the execution state of the first guest OS and the state of the memory used by the first guest OS, the copy of the first guest OS A second guest OS that is generated in the second virtual machine execution environment, and the second guest OS generated in the second virtual machine execution environment is stopped and the state of the second guest OS And a step of storing.

  Japanese Patent Application Laid-Open No. 2006-344025 describes an operation performance data acquisition method (Patent Document 5). The operation performance data acquisition method acquires operation performance data using a computing system. A computing system collects from a computer a plurality of computers that execute various programs, a business server that assigns the programs to the computers, and requests the assigned computers to execute the programs, and operational performance data related to the execution environment of the programs. And a performance monitoring server that monitors the performance of the computer based on the operation performance data. The operational performance data acquisition method includes a step in which a business server includes business information indicating the assigned program and computer identification information for identifying a computer that executes the program, and generates a collection instruction for the program, A step in which the monitoring server reads a collection item corresponding to the business information included in the collection command from the memory and distributes it to a computer having computer identification information; and a step in which the computer collects operational performance data related to the distributed collection item And the performance monitoring server has a step of acquiring operation performance data relating to the collection item from a computer that has collected the operation performance data relating to the collection item by distributing the collection item.

JP 2006-65430 A (Claim 1) JP 2002-32244 A (Claim 1) JP-A-2005-115751 (Claim 1) JP 2006-37763 A (Claim 1) JP 2006-344025 A (Claim 1)

  The conventional operation management apparatus has the following problems.

  As a first problem, in the case of performance information in which the fluctuation range of the load on the guest computer changes when the resource allocation is changed on the host computer, the conventional operation management apparatus expects the change of the resource allocation. There was a problem that an unexpected threshold exceeded (error detection of failure) occurred. In addition, there is a problem that an appropriate monitoring threshold cannot be set for performance information whose fluctuation range is dynamically changed.

  As a second problem, in the conventional operation management apparatus, since the load on the guest computer is detected on the host computer, the utilization rate of the resources allocated to the guest computer can be detected accurately, but actually on the guest computer. In other words, it is difficult to accurately perform correlation analysis with other elements such as AP on the guest computer.

  For example, the CPU load is generally calculated as an average value of detection time intervals. On the host computer, the value can be detected accurately based on all processing times. However, when CPU resources are not allocated to the guest computer, the time information counted on the guest computer is different from the host computer (thinned out). Time information). For this reason, when the AP performs interrupt processing based on the interval time on the guest computer, the abnormal state of the load that can be detected that can be detected on the host computer and the processing of the AP that is processed according to the load that can be detected on the guest computer Even if the messages are compared, it may not be possible to accurately determine the failure.

  The subject of this invention is providing the operation management apparatus which can solve said problem.

  The operation management apparatus of the present invention includes a performance information collection unit, a performance abnormality analysis unit, a resource information collection unit, and a resource allocation analysis unit. The performance information collection unit collects information representing the load performance of the guest computer virtually realized on the host computer as a value, and outputs it as performance information. The performance abnormality analysis unit receives the performance information, compares the value represented by the performance information with a threshold value, and determines that the load on the guest computer is abnormal, outputs an abnormality message indicating that fact. The resource information collection unit collects information indicating the number of resources allocated to the guest computer on the host computer and outputs it as resource information. The resource allocation analysis unit receives resource information and performance information, refers to adjustment data that associates a plurality of natural numbers and adjustment factors, and selects a selection adjustment factor corresponding to the number represented by the resource information from the plurality of adjustment factors. The value represented by the performance information is adjusted according to the selection adjustment magnification, and is output to the performance abnormality analysis unit as performance information.

  Since the conventional operation management apparatus cannot grasp such a change in resource allocation, a fault is erroneously detected. In the operation management apparatus of the present invention, since a change in resource allocation can be grasped, such erroneous detection is suppressed, and accurate failure determination can be performed. In addition, the threshold can be easily set because the determination can be made with a certain threshold regardless of the change in allocation. From this, the first problem can be solved.

  Further, in the operation management apparatus of the present invention, since abnormality determination is performed from information adjusted based on the performance information detected on the guest computer, abnormal states of other elements are appropriately determined according to a state that can be detected on the guest computer. Correlation analysis can be performed. From this, the second problem can be solved.

  Hereinafter, an operation management apparatus according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

(First embodiment)
[Constitution]
FIG. 1 is a diagram for explaining the configuration of an operation management apparatus according to an embodiment of the present invention. FIG. 2 is a block diagram showing the configuration of the operation management apparatus according to the embodiment of the present invention.

  As shown in FIG. 1, the operation management apparatus according to the embodiment of the present invention includes a performance information collection unit 1, a performance abnormality analysis unit 2, a failure analysis unit 3, and an administrator dialogue unit 4.

  The performance abnormality analysis unit 2, the failure analysis unit 3, and the administrator dialogue unit 4 are computer programs provided in the management manager 200, which is a computer, and executed on the computer. The performance information collection unit 1 is a computer program that is provided in the guest computer 110 that is virtually realized on the host computer 100, which is a computer, and is executed by the computer. The guest computer 110 is generated by software.

  The performance information collection unit 1 collects information representing the load performance of the guest computer 110 as a value and outputs it as performance information. The performance abnormality analysis unit 2 has a first function that receives performance information from the performance information collection unit 1 and performs abnormality determination described below. The performance abnormality analysis unit 2 (first function) receives the performance information, compares the value represented by the performance information with a threshold value, and determines whether there is an abnormality in the load on the guest computer 110 based on the comparison result. Analyze. Here, when the value represented by the performance information exceeds the threshold (when the threshold is exceeded), the performance abnormality analysis unit 2 determines that there is an abnormality in the load on the guest computer 110, and an abnormality message indicating that fact Is output (abnormal judgment). The failure analysis unit 3 analyzes whether or not there is a failure in the entire system according to the occurrence order and combination of the abnormal messages, and when determining that there is a failure in the entire system, generates failure state information indicating that (failure Judgment). In this case, the failure state information is notified to the administrator via the administrator dialogue unit 4. The manager dialogue unit 4 presents the fault state information to the manager and receives instructions from the manager.

  As shown in FIG. 2, the operation management apparatus according to the embodiment of the present invention further includes a resource information collection unit 5 and a resource allocation analysis unit 6. The resource information collection unit 5 is a computer program provided in the host computer 100 and executed by a computer (host computer 100). The resource allocation analysis unit 6 is a computer program provided in the management manager 200 and executed by a computer (management manager 200).

  The resource information collection unit 5 collects information indicating the number of resources allocated to the guest computer 110 on the host computer 100 and outputs it as resource information. The resource allocation analysis unit 6 holds adjustment data (see FIG. 5) that associates a plurality of natural numbers and adjustment magnifications. The resource allocation analysis unit 6 receives resource information and performance information. At this time, the resource allocation analysis unit 6 refers to the adjustment data, selects an adjustment magnification (selected adjustment magnification) corresponding to the number represented by the resource information from a plurality of adjustment magnifications, and is represented by the performance information. The value is adjusted according to the selection adjustment magnification, and is output to the performance abnormality analysis unit 2 as the above performance information.

  In addition to the first function, the performance abnormality analysis unit 2 further has a second function that receives performance information from the resource allocation analysis unit 6 and performs the above-described abnormality determination. That is, when the performance abnormality analysis unit 2 receives the performance information from the resource allocation analysis unit 6, the performance abnormality analysis unit 2 compares the value represented by the performance information with a threshold value. At this time, if the value represented by the performance information exceeds the threshold value, it is determined that the guest computer 110 is abnormal, and an abnormal message indicating that fact is output (abnormal determination).

  The performance information collection unit 1, the performance abnormality analysis unit 2, the failure analysis unit 3, the administrator dialogue unit 4, the resource information collection unit 5, and the resource allocation analysis unit 6 are computer programs executed on a computer (operation management device). May be.

[Operation]
With reference to FIG. 2 to FIG. 6, an operation when the load on the guest computer 110 fluctuates due to the resource allocation change on the host computer 100 will be described.

  FIG. 3 is a flowchart showing the operation of the operation management apparatus according to the embodiment of the present invention. FIG. 4 is a diagram for explaining the operation of the resource allocation analysis unit 6 (generation of adjusted performance information). FIG. 5 shows the adjustment data described above, and shows the relationship between the natural number and the adjustment magnification. FIG. 6 shows the failure determination process of FIG.

  Hereinafter, CPU (Central Processing Unit) load on the guest computer 110 is used as performance information, and the number of guest computers {number of virtual machines (VM)) executed on the host computer 100 is used as resource information. Although described, it is not limited to the example.

  The resource allocation analysis unit 6 receives the number of VMs on the host computer 100 (lower part of FIG. 4) as resource information from the resource information collection unit 5 (step 901 in FIG. 3). As the number of VMs increases on the host computer 100, the CPU resource allocated to each guest computer 110 decreases. For example, when the number of VMs is 2, resources on the host computer 100 are allocated by half, and when the number of VMs is 3, it is allocated by one third. For example, in FIG. 4, the number of VMs increases from 2 to 3 from time t1 to time t2, and therefore, a certain amount of resource reduction occurs during the same time in the guest computer 110.

  Further, the resource allocation analysis unit 6 receives the CPU load value (middle part of FIG. 4) on the guest computer 110 as performance information from the performance information collection unit 1 (step 902). In the guest computer 110, since the total CPU resource is decreased from the time t1 to the time t2, the% value of the load for performing processing similar to that at other times is increased.

  The resource allocation analysis unit 6 generates an adjusted CPU load (upper stage in FIG. 4) in which the CPU load value is reduced by a certain percentage only from time t1 to time t2 in accordance with the total resource reduction by the number of VMs (step 4). 903). In this case, the magnification of the performance information adjusted according to the number of VMs varies depending on the capability of the host computer 100. The adjustment data shown in FIG. 5 is an example of such an adjustment magnification. For example, when there are a plurality of CPUs of the host computer 100, there is a high processing capacity up to a certain number of VMs, but as the number increases further. It will drop rapidly. Here, in consideration of the difference between the adjustment magnification when the number of VMs is 2 (selection adjustment magnification in this case) and the adjustment magnification when the number of VMs is 3 (selection adjustment magnification in this case), the CPU load after adjustment Is generated.

  The performance abnormality analysis unit 2 determines whether or not the CPU load is abnormal according to a predetermined threshold value (step 904). Referring to FIG. 6, for example, the failure determination processing 904 determines an abnormality based on a set threshold (step 951), and generates an abnormality message when the threshold is exceeded (step 952-YES) (step 952). 953). When the threshold value is not exceeded (step 952-NO), the process of step 953 is skipped. The failure analysis unit 3 analyzes a combination with an abnormal message other than the performance value such as an abnormal message directly output from the AP, for example, and determines an abnormality of the entire system (step 954).

  If the failure analysis unit 3 determines that there is a failure in the entire system (step 905-YES), it generates failure state information indicating that fact and notifies the administrator via the administrator dialogue unit 4 (step 906). ). Thereafter, the process returns to step 901. If it is determined that there is no failure in the entire system (step 905-NO), the process of step 906 is skipped and the process returns to step 901.

  FIG. 7 is a diagram for explaining the operation (threshold determination) of the performance abnormality analysis unit 2 of the operation management apparatus according to the embodiment of the present invention.

  As shown in FIG. 7, for example, an upper threshold value is set, and when the threshold value is exceeded, an overload is determined and an abnormal message is generated. In the conventional operation management apparatus, since the abnormality determination is performed using the CPU load (lower part in FIG. 7) that has fluctuated due to the resource allocation, the threshold is exceeded between the time t1 and the time t2 when the CPU resource is reduced. On the other hand, in the operation management apparatus according to the embodiment of the present invention, the CPU load after adjustment (upper part of FIG. 7) is adjusted according to the allocated resource amount, and the threshold is not exceeded. .

  In such a CPU resource allocation change, the processing actually performed on the guest computer 110 is not changed at all, and the scale of the detected value is changed. Further, since this increase in load returns to normal by the allocation control on the host computer 100, it is not necessary to present the increase in load during this period to the administrator as a failure.

  Since the conventional operation management apparatus cannot grasp such a change in resource allocation, a fault is erroneously detected. In the operation management apparatus according to the embodiment of the present invention, since a change in resource allocation can be grasped, such erroneous detection can be suppressed and accurate failure determination can be performed. In addition, the threshold can be easily set because the determination can be made with a certain threshold regardless of the change in allocation. From this, the first problem can be solved.

  Further, in the operation management apparatus according to the embodiment of the present invention, since abnormality determination is performed from information adjusted based on the performance information detected on the guest computer 110, other elements are detected according to the state that can be detected on the guest computer 110. Correlation analysis can be performed appropriately with abnormal conditions. From this, the second problem can be solved.

(Second Embodiment)
In the second embodiment, descriptions overlapping with those in the first embodiment are omitted.

  As a third problem following the first and second problems described above, in the case of performance information that cannot be detected on the guest computer even if the resource allocation is changed on the host computer, Even if the resources required for the guest computer are not allocated due to an increase in the load on the host computer or the like, there is a problem that a failure is not detected because the performance is not abnormal on the guest computer.

  In order to explain the above problem and this embodiment together, referring to FIGS. 8 and 9, when the resource allocation on the host computer 100 is changed, the load on the guest computer 110 is not affected. Will be described.

  FIG. 8 is a diagram for explaining the operation of the resource allocation analysis unit 6 (generation of adjusted performance information). FIG. 9 is a diagram for explaining the operation (threshold value determination) of the performance abnormality analysis unit 2.

  Similar to the operation described with reference to FIGS. 8 and 9, the resource allocation analysis unit 6 receives and adjusts the number of VMs (lower part of FIG. 8) as resource information and the CPU load (middle part of FIG. 8) as performance information. A later CPU load (the upper part of FIG. 8) is generated. In this case, the VM number increases from 2 to 4 from time t3 to time t4, but there is no particular change in the peak value of the CPU load on the guest computer 110. This is the case where the CPU clock number is used to control the CPU resource. On the guest computer 110, the total of the given CPU resource does not seem to change, but the time to which the CPU is allocated decreases. Will be. The processing actually performed on the guest computer 110 is delayed because no CPU time is allocated, but the performance information collecting unit 1 on the guest computer 110 cannot detect this.

  As shown in FIG. 9, in the abnormality determination of the conventional operation management apparatus (lower part of FIG. 9), no change is observed in the peak amount of the load, so that the abnormality cannot be detected. In the abnormality determination by the CPU load after adjustment in the present invention (upper part of FIG. 9), it can be detected that the processing amount from time t3 to time t4 is reduced. For example, by providing a lower limit threshold value for peak load, a failure such as processing failure due to insufficient resource allocation can be presented to the administrator. From this, the third problem can be solved.

  That is, in the first embodiment, the performance abnormality analysis unit 2 (first and second functions) has an abnormality in the load on the guest computer 110 when the value represented by the performance information exceeds the threshold (upper limit threshold). And has first and second functions for outputting an abnormal message indicating that. In contrast, in the second embodiment, the performance abnormality analysis unit 2 determines that there is an abnormality in the load on the guest computer 110 when the value represented by the performance information is less than the lower threshold that is lower than the upper threshold. And a third function for outputting an abnormal message indicating that. Thereby, the third problem can be solved.

  Further, the CPU load of the guest OS can be corrected based on the ratio of the CPU resources allocated to each guest OS instead of the number of VMs on the host computer 100. In this case, since it usually does not change dynamically, it can be realized by monitoring an actual CPU clock that is not monitored.

  In addition to the CPU load, the present invention can also be applied to performance information monitoring for all resources on the guest computer 110.

  The operation management apparatus according to the first and second embodiments of the present invention has been described according to the case where the guest computer 110 is generated by software. However, as another embodiment, as shown in FIG. The IO virtualization computer 400 in which functional components can be rearranged in terms of hardware may be used. The IO virtualization computer 400 corresponds to the host computer 100 described above. The IO virtualization computer 400 (host computer) includes a processing unit 310 corresponding to the guest computer 110, a memory 320, an HDD 330, an IO allocation control unit 300 connected to the processing unit 310, the memory 320, and the HDD 330. It is equipped with. In this case, since the resource information collecting unit 5 cannot be easily incorporated in the host computer, the resource information collecting for remotely monitoring the processing unit 310, the memory 320, and the HDD 330 via the IO allocation control unit 300 on the monitoring manager 200. By having the part 5, the present invention can be realized.

It is a figure for demonstrating the structure of the operation management apparatus by embodiment of this invention (1st, 2nd embodiment). It is a block diagram which shows the structure of the operation management apparatus by embodiment of this invention (1st, 2nd embodiment). It is a flowchart which shows operation | movement of the operation management apparatus by embodiment of this invention (1st, 2nd embodiment). It is a figure for demonstrating operation | movement (generation of the performance information after adjustment) of the resource allocation analysis part 6 of the operation management apparatus by embodiment of this invention (1st Embodiment). The adjustment data in embodiment of this invention is shown, and the relationship between a natural number and adjustment magnification is shown (1st, 2nd embodiment). The failure determination process of FIG. 3 is shown (first embodiment). It is a figure for demonstrating operation | movement (threshold determination) of the performance abnormality analysis part 2 of the operation management apparatus by embodiment of this invention (1st, 2nd embodiment). It is a figure for demonstrating operation | movement (generation of the performance information after adjustment) of the resource allocation analysis part 6 of the operation management apparatus by embodiment of this invention (2nd Embodiment). It is a figure for demonstrating operation | movement (threshold determination) of the performance abnormality analysis part 2 of the operation management apparatus by embodiment of this invention (2nd Embodiment). It is a block diagram which shows another structure of the operation management apparatus by embodiment of this invention (1st, 2nd embodiment).

Explanation of symbols

1 Performance information collection unit,
2 Performance abnormality analysis part,
3 Failure analysis department,
4 Administrator Dialogue Department,
5 Resource Information Collection Department,
6 Resource Allocation Analysis Department,
100 host computer,
110 guest computers,
200 monitoring manager,

Claims (11)

A performance information collection unit that collects information representing the load performance of the guest computer virtually realized on the host computer, and outputs it as performance information;
When the performance information output from the performance information collection unit is received and the value represented by the performance information from the performance information collection unit exceeds a threshold, it is determined that there is an abnormality in the load on the guest computer. , A performance abnormality analysis unit that outputs an abnormality message indicating that,
A resource information collection unit that collects information representing the number of guest computers virtually realized on the host computer and outputs the information as resource information;
Receiving the resource information output from the resource information collection unit and the performance information output from the performance information collection unit, referring to adjustment data associating a plurality of natural numbers and adjustment magnifications, the plurality of adjustment magnifications The selection adjustment magnification corresponding to the number represented by the resource information from the resource information collection unit is selected, and the value represented by the performance information from the performance information collection unit is adjusted according to the selection adjustment magnification, A resource allocation analysis unit that outputs to the performance abnormality analysis unit as the performance information,
The performance abnormality analysis unit is
When the performance information output from the resource allocation analysis unit is received and the value represented by the performance information from the resource allocation analysis unit exceeds the threshold, it is determined that there is an abnormality in the load on the guest computer And an operation management apparatus that outputs the abnormal message indicating the fact. The performance abnormality analysis unit is
When the value represented by the performance information is less than a lower threshold that is lower than the upper threshold that is the threshold, it is determined that there is an abnormality in the load on the guest computer, and the abnormality message indicating that is output. The operation management apparatus according to 1. Analyzing whether or not there is a failure in the entire system by the combination of the abnormal messages, and when determining that there is a failure in the entire system, a failure analysis unit that generates failure state information indicating that,
The operation management apparatus according to claim 1, further comprising an administrator dialogue unit that presents the failure state information to an administrator. The guest computer is generated by software,
The operation management apparatus according to claim 1. The guest computer is configured in hardware.
The operation management apparatus according to claim 1. A method performed by a computer,
(A) When the performance information that represents the load performance of the guest computer that is virtually realized on the host computer is received and the value represented by the performance information exceeds a threshold value, the load on the guest computer is abnormal. A step of determining that there is an error message and outputting an abnormal message to that effect;
(B) receiving the resource information indicating the number of the guest computers virtually realized on the host computer and the performance information, and referring to the adjustment data associating a plurality of natural numbers and adjustment factors, Selecting a selection adjustment magnification corresponding to the number represented by the resource information from among the adjustment magnifications, adjusting a value represented by the performance information according to the selection adjustment magnification, and outputting as the performance information;
(C) If the value represented by the performance information from step (b) exceeds the threshold, it is determined that there is an abnormality in the load on the guest computer, and the abnormality message indicating that is output An operation management method comprising the steps of: The steps (a) and (c) include:
When the value represented by the performance information is less than a lower threshold value that is lower than the upper threshold value that is the threshold value, it is determined that there is an abnormality in the load on the guest computer, and the abnormal message indicating that is output. The operation management method according to claim 6. (D) analyzing whether or not there is a failure in the entire system by the combination of the abnormal messages, and when determining that there is a failure in the entire system, generating failure state information indicating that;
The operation management method according to claim 6 or 7, further comprising: (e) presenting the failure state information to an administrator. (A) When the performance information that represents the load performance of the guest computer that is virtually realized on the host computer is received and the value represented by the performance information exceeds a threshold value, the load on the guest computer is abnormal. A step of determining that there is an error message and outputting an abnormal message to that effect;
(B) receiving the resource information indicating the number of the guest computers virtually realized on the host computer and the performance information, and referring to the adjustment data associating a plurality of natural numbers and adjustment factors, Selecting a selection adjustment magnification corresponding to the number represented by the resource information from among the adjustment magnifications, adjusting a value represented by the performance information according to the selection adjustment magnification, and outputting as the performance information;
(C) If the value represented by the performance information from step (b) exceeds the threshold, it is determined that there is an abnormality in the load on the guest computer, and the abnormality message indicating that is output And a computer program for causing a computer to execute each step. The steps (a) and (c) include:
When the value represented by the performance information is less than a lower threshold that is lower than the upper threshold that is the threshold, determining that there is an abnormality in the load on the guest computer, and outputting the abnormality message indicating that,
The computer program according to claim 9, further comprising: (D) analyzing whether or not there is a failure in the entire system by the combination of the abnormal messages, and when determining that there is a failure in the entire system, generating failure state information indicating that;
The computer program according to claim 9 or 10, further causing the computer to execute each step of (e) presenting the failure state information to an administrator.

Images