JPH04107644A - On-line information collection system - Google Patents

Abstract

PURPOSE:To omit the correction of an on-line program and to reduce the working needed for solution of a bottleneck of performance by holding an information collection routine address in a supervisor call SVC table. CONSTITUTION:The resident/non-resident bits, the SVC routine addresses, and the information collection routine addresses are registered in an SVC table 19 for each SVC No. An information collection routine start monitoring part 12 retrieves the information collection routine address out of the table 19 with an SVC No used as a key when a state flag 17 is kept ON (measurement of performance allowed) and also a task ID is stored in a task ID list table 18. Then the part 12 starts an information collection processing program 15 of the SVC. Thus it is possible to obtain the raw data on a concrete on-line with no correction of an on-line program and to reduce the working needed for solution of a bottleneck of performance.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention is mainly aimed at elucidating the causes of abnormalities such as decreased response/turnaround that occur in computer systems of online systems, in an environment where online services are continued. Concerning an online information collection method that enables the actual measurement and analysis of necessary data.

[Prior Art] As conventional online information collection methods, information collection methods using log data, information collection methods using online program modification, etc. are known.

The log information collection method is a method of collecting resource usage status logs such as the CPU, terminal responses, and statistical logs such as throughput at regular intervals for all tasks during the entire online period. In addition, the online program modification method is
A performance analysis monthly information collection routine, sometimes called an information collection trap, is embedded in the online program to collect and analyze information, and once the online analysis is complete,
This method reverts to the original online program.

[Problem to be solved by the invention]

In the conventional method described above, in the log information collection method, the information to be collected is
/Limited to macro phenomenon information (information for equipment renewal) such as memory usage rate and terminal response. Therefore, specific information (such as the name of the task/processing routine that caused the increase in CPU processing time) cannot be collected until the true cause of the problem that has appeared as a phenomenon is clarified and directly linked to countermeasures. As a result, cause estimation, etc. must be performed in a situation where there is no basis data, and the required operation is large and takes a long time to uncover the performance bottleneck.

Furthermore, in the online program modification method, in order to embed or remove the information collection routine in the online program, the program must be replaced, and the online service must be interrupted each time. In addition, since it is not possible to embed information collection routines in all parts of an online program, it is necessary to repeatedly target the parts that are likely to cause problems based on experience, and as with the log information collection method, , the time and operation required to investigate the cause was significant. Furthermore, with the embedding and removal of a large number of information gathering routines, there is also the problem that operational errors are likely to occur and online quality is likely to deteriorate due to the introduction of program bugs.

The present invention aims to solve the problems of the prior art described above, and makes it possible to continue operating online services by eliminating the need to modify online programs.
In addition, the system automatically collects specific performance analysis information while minimizing deterioration of online performance, thereby allowing early measures to be taken against performance bottlenecks.

[Means to solve the problem]

In order to achieve the above object, the present invention transfers control between programs using SVC (supervisor cal).
l) In the online processing computer system by issuing/returning and issuing RTJN/STOP macros,
A status flag indicating whether or not online processing performance is being measured;
Task ID that holds the measurement target task ID list
An SVC that has a list table and manages resident/nonresident and SVC processing routine addresses for each SVC.
The table also holds an information collection routine address, and when an SVC is issued from an online task, when the status flag indicates that performance is being measured and the task ID exists in the task ID list table, the SVC table is stored. The information collection program is started based on the information collection routine address, and after the information collection processing is completed, the SVC processing routine address is
This is to start the processing program.

[Function] In the present invention, when an SVC is issued from an online task, if the computer is measuring online processing performance and the measurement target task ID is registered, the SVC processing is automatically executed prior to the original SVC processing. performs information collection processing for performance evaluation, then shifts to the original SVC processing,
Otherwise, the original SVC processing is directly performed. This makes it possible to obtain raw data for specific online processing without modifying the online program, and to minimize the impact on online performance.

〔Example〕

An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 shows a functional block diagram of an embodiment of the system of the present invention. In FIG. 1, +1 is the SVC analysis processing section, 12 is the information collection routine startup monitoring section, and 13 is the SVC
These are routine startup parts, and these are C
It constitutes part of the processing function of the PU. 14 is an online task, and a unique task ID is set for each task.

15 is an information collection processing program, 16 is an SVC processing program, and there are a plurality of these depending on the type of SVC.
Some of them are resident on the main memory, while others are placed on the external memory (non-resident) and loaded into the main memory each time they are needed. , 17 is a performance measurement status flag, 18
19 is a measurement target task ID list table, and 19 is an SVC table, which may be located in either the register group of CPtJ or the main storage device.

Figure 2 shows status flag 17 and task ID list table 1.
8, is a detailed diagram of the SVC table 9.

(a) is a performance measurement state flag 17, which is switched to 0N10FF at the start time and end time of performance measurement during online performance based on instructions from the operator, automatic driving program, etc. In other words, "ON" means performance is being measured (actual performance measurement allowed), and "rOFF" means performance is not being measured (actual performance measurement inhibited).
shows.

(b) is a measurement target task ID list table 18, in which IDs (A, R, etc.) of measurement target tasks designated by the operator, automatic driving program, etc. are stored. The unregistered part is rNULLJ.

(c) is the SVC table 19, in which N class of SVC is
S is expressed as a VC number and indicates the resident/nonresident bit and a pointer to the corresponding SVC processing program for each SVC number.
A VC routine address and an information collection routine address indicating a pointer to the corresponding information collection processing program are registered.

Next, the operation shown in FIG. 1 will be explained. Online task 14
When a certain SVC is issued, the SVC analysis processing unit 11
analyzes the SVC and passes the task ID set for the task and the type of SVC (SVC number) to the information collection routine activation monitoring unit 12.

In the information collection routine startup monitoring unit 12, the status flag 17
, refer to the task ID list table 18, and if the status flag 17 is No (actual performance measurement allowed) and the task ID is stored in the task ID list table 18, information is retrieved from the SVC table 19 using the SVC number as a key. The collection routine address is searched and the information collection processing program 15 of the SVC is started. After the information collection processing program 15 completes the information collection processing, control is passed to the SVC routine starting section 13. On the other hand, status flag 17 is OFF
(actual performance measurement suppression) or status flag 17 is O
Even if N, the relevant task ID is displayed in the task ID list table 18.
If not stored, the information collection routine activation monitoring unit 12 immediately passes control to the SVC routine activation unit 13 without starting the information collection processing program [5].

FIG. 3 shows the processing flow of the information collection routine activation monitoring section 12.

The SVC routine starting unit 13 uses the SVC number as a key to
The SVC routine address is searched from the VC table 19, and the processing program 6 for the SVC is started. SVC
Processing program】6 is the same as before, to execute the relevant SVC process (if it is a non-resident SVC, it will be executed after the SVC is loaded)
, return to the SVC issuer.

Next, a specific example will be explained with reference to FIGS. 4 and 5.

FIG. 4 shows the case of online normal operation, and the status flag 17 is OFF. In this case, from the online task (task ID = A) to SVC (SVCα
When C■) is issued, the SVC routine address is obtained from the SVC table]9 (in this example, it is 1000).
), the processing program of the SVCα is started (■), the original processing of the SVCα is executed (■), and when these are completed, the process returns to the issuer of the SVCα (■).

FIG. 5 shows a case where online performance is being measured, and the status flag ]7 is ON. It is assumed that ID=A is stored in the task ID list table 18 as the task ID to be measured. When an SVCα is issued from an online task to task ID= (■), the information collection routine address is first obtained from the S■C table 19 (5000 in this example) in mj of the original SVCα processing, and the corresponding SVC
Please start α's information collection processing program.
Collect information necessary for performance evaluation of α (■).

When this information collection is completed (■), obtain the SVC routine address from the SVC table 19 (1000 in this example), start the processing program of the SVCα (■), and execute the original processing of the SVCα. (■), returns to the issuer of the SVCα (■). Furthermore, task I
When an SVC is issued from an online task that is not stored in the task ID list table 18, such as when D=B, the SVC is issued even if the status flag 17 is ON.
The original VC processing is directly executed.

Here, in conjunction with the control transfer between the information collection processing program and the information collection processing program, the original control transfer between programs needs to be performed as before. For this reason, although it is omitted in the figure for the sake of simplicity, it is natural that the handover information for control transfer between programs, such as the new and old PSWs and registers, is in the state before the original program is called (the next execution address is the relevant one). It also has a function to save at the execution address of the SVC processing program and revive it to that state after completion.

In addition, in Fig. 5, the information collected by the information collection processing program is written to memory, etc., but if multiple memories are provided and one memory becomes full, it is switched to another free memory. It is also possible to provide so-called multiple buffer control, which flushes out the contents of a full memory to an external storage (file) in parallel with adding new collected information to the memory.

[Advantageous Effects of the Invention 1] As described above, in the present invention, specific online raw data, such as OS kernel management information that cannot be obtained from log information and can clarify the direct cause of performance bottlenecks, can be used without modifying online programs. As a result, the operation required to investigate the cause of performance bottlenecks can be significantly reduced, and the problem can be resolved quickly.

Furthermore, by limiting the collection of performance analysis information to specified tasks and performing it only during a specified period, the impact on online performance can be minimized. The impact on online performance can be further reduced by performing editing and analysis after information collection only in the offline system.

In addition, in a long-term continuous operation system, if the status flag is set to the non-measuring state, there will be no control transfer to the information collection program. It is also possible to replace it with a new function-added board whose quality has been confirmed.

In the future, with the increasing sophistication of online system operations, hardware and software will become larger and more complex, and continuous operation time will tend to expand. The importance of provision is only increasing. The present invention is positioned as a new basic method to meet this need.

[Brief explanation of the drawing]

FIG. 1 is a functional block diagram of an embodiment of the method of the present invention, and FIG.
The diagram shows status flags, task ID list table, and SV
A detailed diagram of the C table, Figure 3 is a processing flow diagram of the information collection routine startup monitoring section, Figure 4 is a diagram showing an example of processing in online normal operation, and Figure 5 is processing in the case of online performance evaluation information collection. It is a figure which shows an example. 11...SVC analysis processing unit, 2...Information collection routine startup monitoring unit, 3...SVC
Routine starting part, 4...Online task, 5...Information collection processing program, 6...SVC processing program, 7...Status flag, 8...Task ID list table, 9...SVC table. Figure 2 (O)

Claims (1)

[Claims] (1) Control transfer between programs is performed using SVC (Super-
visor call) issuance/return and RUN/S
In an online processing computer system performed by issuing a TOP macro, a status flag indicating whether or not online processing performance is being measured;
Task ID that holds the measurement target task ID list
A list table and an SV that holds information collection routine addresses in addition to SVC processing routine addresses for each SVC.
C table is provided, and when an SVC is issued from an online task, if the status flag indicates that performance is being measured and the task ID exists in the task ID list table, the SV
Refer to the C table, start the information collection program based on the information collection routine address, and after the information collection processing is completed, start the information collection program based on the SVC processing routine address.
An online information collection method characterized by starting a VC processing program.