JPH0348946A - Monitor system for system load state - Google Patents

Abstract

PURPOSE:To quickly cope with the deterioration of the system processing performance, etc., by comparing the latest job information with the job information preceding by a unit time and extracting only the common jobs. CONSTITUTION:A stack means 2 extracts the job information, the information on the resources used by the jobs, and the job behavior information out of a system table 1 and sends these information to a comparison means 4. The means 4 compares Sa-1 with Sb-1 of a divided area 3-1 of a stack area 3 and stores only the common job information in the Sa-2. This procedure is repeated to the divided areas 3-2 - 3-n and finally the job information is stored in Sa-(n + 1). Thus it is possible to know the job name that worked continuously for (n) unit times, the job name that used continuously the same file, the job name that outputted continuously the same message, and the job name that repeated the start and the end. As a result, it is possible to detect the cause that gives the wrong load to the system and to quickly cope with the deterioration of the system processing performance, etc.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a system load status monitoring method for monitoring the load status on system resources of a convenitor system.

[Conventional technology] Conventionally, to monitor the load status of a computer system, a static monitoring method was used in which a list of system operation results was output when necessary, and workers analyzed and tracked the results based on the list. was.

[Problems to be Solved by the Invention] As described above, the conventional system load status monitoring method described above mainly relies on manual information analysis. As a result, it is not possible to detect the problem at the time it is occurring, which slows down the pursuit of the cause of the abnormally high system load and problem resolution, and prolongs the system's poor operation, resulting in system sloop. There were drawbacks such as a decrease in processing performance, a decline in processing performance for online operations, and an increase in costs due to manual work.

In addition, there are security concerns in that even if system resources (e.g. memory, CPU, free space on magnetic disks, lines, etc.) are used illegally for a long time by a specific unauthorized job, this cannot be detected. There were also problems.

Note that if threshold control is implemented in various parts of the operating system in an attempt to detect this, the overhead required will conversely deteriorate the performance of the system and increase costs.

The present invention improves these conventional drawbacks.The purpose of the present invention is to dynamically monitor system load status, quickly resolve problems, and effectively prevent deterioration of system processing performance and security problems. An object of the present invention is to provide a system load status monitoring method that can prevent such problems.

[Means for Solving the Problems] The system load status monitoring method of the present invention reads ship information, which is information on the usage status of system resources related to the job being executed, from a system table every unit time. A stack means copies the job information to stack areas divided according to the number of hours, a save means saves the job information copied to the stack area, and a save means saves the latest job information previously copied by the stack means. The comparison means compares the saved job information of one unit time ago, extracts only the common job information, and finally obtains the common job information for a predetermined number of unit times.

[Operation] The stack means reads job information as information on the usage status of system resources related to the job being executed from the system table every unit time, and loads this into stack areas divided according to the number of units of time. Copy to. The saving means saves the job information copied to the stack area, and the comparing means compares the latest job information copied by the stacking means with the job information one unit time ago saved by the saving means to find a common information. Extract only the job information to be used. This process is repeated and performed sequentially for a predetermined number of unit times, and finally common job information is obtained through the predetermined number of unit times.

This common job information is information that has occurred continuously,
By outputting this, the system load status can be dynamically monitored.

[Embodiment] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram of an embodiment of the present invention. In FIG. 1, the system table l includes system resources related to the job being executed, such as memory, CPU, etc. on the computer's main storage and secondary storage (not shown).
Information indicating the usage status of magnetic disk devices, lines, etc. (hereinafter referred to as job information) is stored.

In addition, the stack area 3 includes main storage, secondary storage (
(not shown), which is logically divided into (2n+1) ways, S and -11tllll respectively.
, S, -1 to S, -1, and S b-I to S b-n. In addition, in Figure 1, S, -8 and S, -1 and S a
Although the relationship with -1e+ is illustrated, such a circuit or the like does not originally exist in this table 3, and merely shows the flow of data.

The stack means 2 extracts predetermined information regarding system resources from the system table l, such as job information such as the occurrence number of each job, the memory used, the CPU used, etc. is copied to the stack area 3 (DS--+), and the comparison means 4 is notified of its completion.

Comparison means 4 compares S, -1 and Sl in star 1 querier 3.
, 1 and select only the common data as S a-1*1
The process of storing data in divided areas 3-1 to 3-111”1
1 is sequentially executed, and when the storage in Sa, a*+1 is finally completed, the save means 5 and the file output tool 6 are notified.

Further, the save means 5 saves S, -8 to S of the stack area 3.
S, -1 to S with the same contents as S, -7.

-, it becomes heavy.

The file output tool 6 consists of a program for writing the final result 3, -(ae+) obtained by the comparison means 4, into a log file 7 (accumulation system file) so that it can be retrieved all at once. dump tool 8
reads the final result group from the log file 7, edits it, and sends it to an output device such as a printer or terminal! ! ! It consists of programs displayed in 9.

Note that the log file 7 is a file on a magnetic disk, and the timer 10 is normally provided in a computer system.

Next, the system load status monitoring operation in such a configuration will be specifically explained. First, when the timer 10 issues a notification to the stacking means 2 at time to, the stacking means extracts information about the job, information about the resources used by the job, and information about the behavior of the job from the system table l. Control is transferred to comparison means 4. Information regarding the job includes the job name, job origination number, user name, billing name, program name, etc. Information regarding the resources used by the job includes the CPU, memory, file name, peripheral device, etc. Information about job behavior includes output messages, start/end, normal/
There are distinctions between abnormalities, etc.

When the comparison means 4 is transferred control from the stacking means 2,
First, 5a-1 and Sb-1 of divided area 3-1 are compared and only common job information is stored at 81°.

Next, Sl, , and 311 are compared, and only the common job information is stored in Sl, , . Hereafter, the same procedure is repeated and finally divided area 3. Compare S,-1 and S,-1 of n, store job information in S,-10, and transfer control to save means 5 and file output tool 6. It is assumed that the stack area 3 is cleared to "0" or blank when the system starts operating. Therefore, at this point, S, 1. Since there is no common information between and S m-1, no job information is entered in 5a-t to S, -10, . Next, the saving means 5 saves the comparison means 4
When IJII is transferred from S10. ~S, -7 data S, -□ ~S, 0. Hecobee. Therefore, S,
, , are stored with data at time to.

When the above processing is repeated until time tn in this way, the stack areas S1.1 to S at each time.

, 1 changes as shown in FIG. As can be seen from Fig. 2, at the caution time tx, S a-8
Common job information within n unit time from time tml-111111) to [, is stored in 61.

For example, the following events can be detected based on the data stored in S a -1a□ in this way. In other words, from the information about the job in the system table, we can find out the name of the job that has continued to operate for n units of time, and from the information about the job and the information about the resources used by the job, we can know whether the same file has continued to be used for n units of time. Also, from the information about the job and the information about the behavior related to the job, the name of the job that has continued to issue the same message for the n1th time, the name of the job that has been started/ended repeatedly, and the user name of the job can be found.

When the file output tool 6 receives control from the comparison means 4, the file output tool 6 reads S a −(* e r 3
The data is written and accumulated in the log file 7. System engineers and system operation managers can download data from log file 7 using dump sole 8 at any time! The result can be printed out on an output device 9, such as a printer, for reference. As a report obtained at this time, for example, the following information can be collected as an MI collection. In other words, the content described based on Figure 2 may be printed as is, or the content described based on Figure 2 may be further compared over m times the unit time to obtain a higher concentration (with a longer sampling time). ) data can be printed.

[Effects of the Invention] As explained above, the present invention calls the jib information every unit time and copies it to the stack area, and stores the latest job mW copied to the stack area and the saved job one unit time ago. By comparing the job information and extracting only the common job information, and finally finding the common ship information for a predetermined number of unit times, system resources can be saved from among the ships that continue to reside in the system. It is now possible to dynamically detect at any time things that continue to occupy the system, things that continue to behave abnormally such as sending messages or repeatedly starting and ending, or that are causing an illegal load on the system. This has the effect of being able to quickly detect malfunctions in the system and quickly respond to a drop in system throughput and deterioration in online business processing performance. In addition, problem detection can be automated, reducing costs for workers, etc., and protecting the stable operation of the system from incomplete programs that unnecessarily monopolize system resources or repeat meaningless processing. You can also. Regarding this security, since it is only necessary to add sampling processing for each unit time, there is no need to implement threshold control in various parts of the operating system as in the past, and it is possible to effectively prevent increases in cost and performance degradation. can.

[Brief explanation of drawings]

FIG. 1 is an overall configuration diagram of an embodiment of the present invention, and FIG. 2 is a diagram showing how the contents of the stack area change at each time. In FIG. 1, l...system technology, 2...stack means, 3...stack table, 4...comparison means, 5...save means, 6...file output tool, 7. ...dump tool, 9...output device, 1G...timer.

Claims (1)

[Claims] A stack means for reading job information as information on the usage status of system resources related to the job being executed from a system table every unit time and copying it to a stack area divided according to the number of units of time; and a stack means for copying to the stack area. a save means for saving the job information that has been created;
The latest job information previously copied by the stacking means and the job information of one unit time ago saved by the saving means are compared, and only the common job information is extracted, and finally the job information is stored for a predetermined number of units of time. and a comparison means for determining common job information in the system.