JP2000357099A - Disk input/output method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To shorten a response time by issuing a disk input/output process request when the start time of a disk input/output process is earlier than the start time when an application program which is being actuated is actuated next time. SOLUTION: A next stream reference interface 105 is used to check for which stream identifier a disk input/output scheduler 104 processes a disk input/ output process request in a next time slot. At this time, a disk input/output process start time reference interface is used to check the time when the disk input/output process request for the stream identifier to be inputted and outputted is processed. It is checked whether or not the checked time is later than the time when the application program 102 is started next time. When the start time of the disk input/output process is earlier, the application program determines data for which the input/output process request is made and issues a disk input/output process request.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for recording a plurality of continuous media data on an information recording device such as a disk drive, and simultaneously outputting a plurality of continuous media data from the information recording device.
The present invention relates to a disk input / output method for inputting / outputting data.

[0002]

2. Description of the Related Art In a device for simultaneously reading a plurality of continuous media data from a disk device and transmitting the read data to a terminal, such as a video server in a video-on-demand service, the number of terminals that can be simultaneously serviced is increased. In addition, there is a demand for a reduction in response time to requests from terminals. In the conventional technology, continuous media data is recorded in stripes on a plurality of disk devices, and the phase of a time slot table indicating when and which disk input / output processing is to be performed for each continuous media data is shifted for each disk. By doing so, a function is provided to reduce the time from issuance of a disk input / output processing request to the start of the processing.

For example, there is a video library system as disclosed in Japanese Patent Application Laid-Open No. 7-123398 as a multiplex reading device of this type of continuous media data. FIG. 14 shows the configuration of this video library system. The video library system compresses and stores continuous media data in a disk device DK141 (14).
01) is transmitted through a plurality of buffer memories BM1401 (1403),... Via a switch SW141 (1402).
BM1432 (1409), all of which are controlled by the controller CN141 (1415).

In this multiplex reading device, the control device CN141 manages a time slot table shown in FIG. In the example according to FIGS. 14 and 15, the terminal TE1
,.., TS1532 are assigned to 401,..., TE1432. The control device CN141 has a time slot TS
The continuous media that the TE 1401 is playing at 1501
A data segment (data unit to be read in one time slot) is read, and the read segment is SW
141 to the buffer BM 1401. The segment in the buffer BM1401 is the decoder DE14.
01 and is displayed on the display device TV 1401 (1502). Similarly, the time slot TS
At 1502, the segment of the continuous media data being reproduced by the terminal TE 1402 is read, and the read segment is stored in the buffer BM 1402 by the SW 141.
Is forwarded to The segments in the buffer BM1402 are decoded by the decoder DE1402 and displayed on the display device TV1402 (1503). This is performed up to time slot TS1532. The above is the operation of one cycle of the multiplex reading device, and the control device CN141 repeats this.

[0005] A terminal TE1401,
…, TE1431 is already connected and terminal T
When E1432 is newly connected, the control device CN141
In order to allocate a time slot to the terminal TE1432, it first searches for an empty time slot. An empty time slot is an unassigned time slot that does not perform input / output to the disk device in that time slot. When the time slot table has 32 time slots and the number of terminals connected to the multiplex reading apparatus at a certain time is n, the number of free time slots at that time is (32-n) Individual. The response time from when the newly connected terminal TE1432 requests playback of continuous media data to when it is played back is that the number of empty time slots is one and that the next idle time The worst case occurs when the number of time slots up to the time slot is equal to the number of time slots in one cycle of the time slot table.

In order to shorten the response time, the conventional multiplex reading apparatus uses the following method.

(1) Continuous media data is recorded in a stripe form.

(2) A time slot table is created for each disk and its phase is shifted.

(3) Data of the first few seconds of continuous media data is cut.

First, as (1), the number of disks of the multiplex reading device is set to a plurality, and continuous media data is recorded in a stripe shape.

Next, as (2), a time slot table is created for each disk. The time slot tables for each disk shift their phases relative to each other so that the time slots for each disk assigned to the same terminal are distributed.

FIG. 16 shows a timing chart of the multiplex reading device when there are four disk devices. 4
Each of the disk devices has a time slot table. The time slot table of the first disk device is TST161 (1601), and the time slot table of the second disk device is TST162.
(1602) The time slot table of the third disk device is TST163 (1603), and the time slot table of the fourth disk device is TST1.
64 (1604).

[0013] Time slot table TST161
In the time slot TS16101, the terminal TE14
01 processes the disk input / output processing request issued. At the same time, in the time slot TS16225 of the time slot table TST162, the terminal TE142
5 processes the disk input / output processing request issued. Similarly, in the time slot TS16317 of the time slot table TST163, the terminal TE1417 transmits the time slot T of the time slot table TST.
In S16409, the disk I / O processing request issued by the terminal TE1409 is processed. The following time slot TS16102 of the time slot table TST161
Then, a disk input / output processing request issued by the terminal TE1402 is processed. At the same time, in the time slot TS16226 of the time slot table TST162, the terminal TE1426 sets the time slot table T
In time slot TS16318 of ST163, terminal TE1418 receives time slot table TST1.
In 64 time slots TS16410, terminal TE1
The disk I / O processing request issued by 410 is processed.
Thereafter, it is repeated according to the time slot table. The processed disk input / output request is reproduced in each terminal. The state of reproduction is shown at 1605 to 1607 in FIG.

The response time of the newly connected terminal TE1432 is equal to the time from the current time slot to the free time slot. At this time, since the empty time slots are out of phase between the disks, TE1432
A free time slot for a visit is made every eight time slots. That is, the worst response time when the terminal TE1432 is added becomes equal to or less than eight time slots. The response time is reduced to a quarter as compared with the example of FIG.

However, the newly connected terminal TE1432
Does not always exist in the disk device corresponding to the empty time slot found above. Then, as (3), the data from the head data of the continuous media data to the portion recorded on the disk device corresponding to the empty time slot is cut. This cut,
You cannot play back continuous media data,
Conversely, the response time becomes faster.

[0016]

The above prior art solves the problem that the response time becomes longer in the periodic disk input / output processing using the time slot table, and shortens the response time. it can. However, when the operating system provides a function of multiplex reading of continuous media data, the problem that response time increases cannot be solved. That is, by using the conventional technology, when a multiplex reading function of continuous media data is realized on a single processor computer system, the timing at which an application program issues a disk I / O processing request and the operating system There is a problem that a shift occurs in processing timing, and as a result, a response time increases.

The problem of this timing shift will be described below with reference to the timing chart of FIG. However, for simplicity, the number of disk devices is one.

First, four application programs AP171,..., AP174 operate on the processor, and each read one piece of continuous media data.
The control device allocates time slots TS171,..., TS174 to each application program.
The operating system is the application
Performs CPU scheduling for the program. ST171 in FIG. 17 is an example of the scheduling table.
An application program that reads continuous media data is periodically started by the operating system at the same cycle as that of the time slot table. Then, the application program issues a request to read a certain number of segments in each cycle. In the timing chart of FIG. 17, the application AP 174 is activated at time T1, and issues a continuous media data read request at time TR1. Next, at time T
2 starts the application program AP173, and issues a read request for continuous media data at time TR2. Similarly, the application program AP
Continue to 171. In parallel with the above, the control device sends a read request from each application program in time
Process according to the slot table. The request of the application program AP174 is time slot T
In S174, the application program AP173
Is in time slot TS173, the request for application program AP172 is in time slot TS172, and the request for application program A is
The request of P171 is the time slot T in the next cycle.
The processing is performed in S171 (1701).

FIG. 17 shows the worst response time WRT17AP174 for the application program AP174.
showed that. When the application program AP 174 is, for example, a video server, a request for skipping continuous media data from a terminal may be issued at a time T5 immediately after the application program AP 174 issues a disk input / output request. There is. In this case, the issuance of the skip request is delayed until time TR6, and the response time at that time becomes the worst.

On the other hand, the scheduling shown in FIG.
The worst response time in the example of the table ST181 and the same time slot table TST171 as in the example of FIG. 17 is about one half of the worst response time in the example of FIG. The reason is shown below. In the example of FIG. 17, the response time results from two types of waiting time. The first waiting time is a waiting time from when the application program AP 174 issues a request to when it is processed, and corresponds to a maximum of about one period of the time slot table. The second waiting time is a waiting time until the application program is scheduled, and corresponds to one cycle of the time slot table. On the other hand, in the example of FIG. 18, the response time is generated from one type of waiting time. It is the waiting time before an application program is scheduled. The reason why the worst response time WRT17AP 174 in FIG. 17 is longer than the worst response time WRT18AP 174 in FIG. 18 is that the waiting time from when an application program issues a request to when it is processed is due to the former response time. This is because it occurs.

As can be seen from the comparison between FIG. 17 and FIG. 18, the above-mentioned waiting time is based on the scheduling table of the application program and the time table of the disk drive.
Caused by a time lag between the slot table. While the time slot table and the scheduling table are synchronized as shown in FIG. 18, they are not synchronized in FIG. Therefore, if the time slot table and the scheduling table are not synchronized, the response time increases.

A main object of the present invention is to make it possible to reduce the response time in an apparatus that provides general-purpose multiplex reading that is not specialized for a specific application such as a video library system.

[0023]

In order to solve the above-mentioned problems, the present invention comprises a plurality of disk devices for recording a plurality of continuous media data, and when and which continuous media is stored.
A computer system having a time slot table indicating whether to perform disk input / output processing on data and performing disk input / output processing based on the time slot table has the following means.

First, the application program specifies the identifier of the continuous media data to the disk input / output scheduler, and determines when the disk input / output processing for the specified continuous media data is performed by the disk input / output scheduler. It has a processing unit for calculating, and has means for the application program to refer to the start time.

Secondly, there is provided a processing unit for calculating the time at which the running application program is next started,
The application program has means for referring to the start time.

Third, there is provided a processing unit for comparing the start time referred to in the first and the start time referred to in the second, and if the start time referred to in the first is earlier, It has means for issuing an input / output processing request.

[0027]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described with reference to FIGS.

The disk input / output method according to the present invention is realized on a computer system. FIG. 12 shows the hardware configuration of the computer system. The computer system includes a processor (1201) for executing a program, a main memory (1202) for temporarily recording the program, and a plurality of disk devices (108, 12) for recording continuous media data.
05), an input / output controller (1204) for controlling the disk device, and a timer (110) for giving a trigger for CPU scheduling, and these are connected by a bus (1203).

FIG. 1 shows a processing flow of a disk input / output processing request for realizing the disk input / output method of the present invention. Details of each process will be described later with reference to other drawings.

The computer system includes a plurality of continuous media
One disk device (10) for recording data (109)
8) are connected and their continuous media data (10
There are a plurality of application programs (102) that periodically issue disk I / O processing requests to 9).

The CPU scheduler (101) started by the timer interrupt starts these application programs (102) periodically. The CPU scheduler starts at the same cycle as those application programs (102). The same period is used to simplify the description, and these may be different. In that case, it will be described at the end of the description of this embodiment.

Application program (102)
Determines the timing of issuing a disk I / O processing request using the next stream reference interface (105) so as to minimize the waiting time from issuance of the disk I / O processing request to processing.

The CPU scheduler (101) periodically starts the disk input / output scheduler (104).
The startup cycle of the disk input / output scheduler (104) is made equal to the startup cycle of the application program (102). FIG. 13 shows an example of the scheduling table of the CPU and an example of the scheduling table at this time. Within one time slot (1306), the CPU
The scheduler has activated the application programs (1301,..., 1304) and the disk input / output scheduler (1305).

The disk input / output scheduler (104) passes the disk input / output processing request issued by the application program (102) to the disk input / output processing program (106) in accordance with a time slot table (107) described later. The disk input / output processing program (106) processes the received disk input / output processing request.

A disk input / output scheduler (104),
The disk input / output module (103) collectively refers to programs, data, and interfaces related to disk input / output processing including the time slot table (107), the next stream reference interface (105), and the disk input / output processing program (106). ).

The data used by the disk input / output module (103) will be described with reference to FIGS.

FIG. 2 shows a stream identifier management table (2
01). Stream identifiers are assigned to continuous media data that is periodically input / output from the disk. Whether or not a stream identifier has been assigned is written in the availability field (203) of the stream identifier management table (201). If a stream identifier has been assigned, enter "in use"; if not, enter "unused".

FIG. 3 shows the time slot table (10
7) is an example. Disk I / O scheduler (10
The time from the start time of 4) to the next start of the disk input / output scheduler (104) is called a time slot. Disk I / O scheduler (1
04) when the time slot is activated for the first time is set to 1
And Each time the disk input / output scheduler (104) is activated, the value of the time slot increases by one. FIG.
In the example, there are 1 to 8 time slots. After time slot 8, return to time slot 1. The time slot table (107) describes which stream identifier should be used to process a disk input / output processing request in each time slot. A certain time
If there is no stream identifier to be processed in the slot, the stream identifier field corresponding to the time slot in the time slot table (303)
Write X in In this embodiment, the number of stream identifiers in the stream identifier management table (201)
The number of time slots in the time slot table (107) is matched.

FIG. 4 shows a next time slot management variable (40).
1). The value of the time slot next to the current time slot is recorded.

FIG. 5 shows a disk input / output command list.
It is an example of a header (501). The disk I / O processing request issued by the application program (102) is held in the disk I / O command list header (501) until it is processed by the disk I / O scheduler (104). The disk input / output processing request is converted into a disk input / output command (504), and the disk input / output command (504) is connected to the disk input / output command pointer (503) of the disk input / output command list header (501). You. In the example of FIG. 5, the disk I / O processing request is continuous media data (10
Only the read request for 9) is considered. At this time, the disk input / output command (504) is composed of three fields. The first field is the read address (505). The read address (505) is a logical address on the disk where data to be read exists. The second field is the lead size (506). The read size (506) is the size of data to be read. The third field is the next command pointer (507). A plurality of disk input / output commands are connected in a list by the next command pointer (507). This list is called a disk input / output command list (508). A plurality of disk input / output commands are stored in a FIFO (File
st In First Out). The new disk I / O command is connected to the end of the disk I / O command list (508).

The disk input / output scheduler (104) processes one disk input / output command in one time slot. Therefore, one disk I / O command (5
The read size (505) in 04) must be large enough to complete its processing within one time slot. If the application program (10
2) The disk I / O processing request issued by
If the processing cannot be completed in the slot, the disk I / O processing request is divided into a plurality of disk I / O commands and processed over a plurality of time slots.

In the example of FIG. 5, disk input / output processing requests are managed separately for each stream identifier (502).
This is an example, and disk input / output commands for all stream identifiers may be managed collectively in one list. In this case, the stream input / output command may include the stream identifier as a new field.

The disk input / output command list shown in FIG.
The header (501) has a stream identifier field (5
X) exists in the value of 02). As described in FIG. 3, the stream identifier X means that there is no continuous media data (109) to be processed in that time slot. In this case, the disk input / output scheduler (104) processes disk input / output processing requests other than periodic disk input / output processing requests. This is called an aperiodic disk input / output processing request. The aperiodic disk input / output processing request is converted into a disk input / output command and connected to the disk input / output command list corresponding to the stream identifier X in the disk input / output command list header (501).

In the above description, only a read request is considered as a disk input / output processing request, but it can be easily extended to a write request.

An example of the operation of the application program (102) for synchronizing with the time slot table will be described with reference to FIG. In the example of FIG. 6, the application program (102) periodically issues a disk input / output processing request for one piece of continuous media data (109). At this time, in order to synchronize the timing at which the application program issues the disk I / O processing request with the time slot in which the disk I / O processing request is processed, the application program uses the next stream reference interface described later.

First, at step 602, the application program (102) periodically opens continuous media data (109) to be input / output.

In the next step 603, using the periodic input / output request interface, the disk input / output scheduler (102) periodically processes the disk input / output processing request for the opened continuous media data (109). Request to the output module (103). Details of the periodic input / output request interface will be described later. The periodic I / O request interface returns a stream identifier to the application program (102).
The value is recorded in the variable's own stream identifier. If the value of the variable's own stream identifier is NULL (step 60)
4) means that the disk input / output module cannot periodically process any more continuous media data.
At this time, the application program (102) ends the processing (step 605).

In step 606, the disk input / output scheduler (104) checks the disk input / output processing request for which stream identifier in the next time slot using the next stream reference interface (105). The stream identifier processed in the next time slot is stored in a variable next stream identifier. Details of the next stream reference interface (105) will be described later.

In step 607, the value of the variable next stream identifier is compared with the value of the variable own stream identifier. If they are different, even if the application program issues a disk I / O processing request, the request is not immediately processed by the disk I / O module. Then, in order to wait until a time slot for processing the own stream identifier comes around, CPU handoff is first performed (step 608). CPU handoff is an interface that relinquishes the rest of the CPU time allocated to the application program itself. If the OS does not have an interface for CPU handoff, the application program is allocated by arbitrary processing.
You just have to spend the rest of your CPU time. When this application program is started again, step 607 is executed.
repeat.

At step 609, the application program determines the data for which an input / output processing request is to be made,
A disk input / output processing request is issued to the disk input / output module (103) using the input / output processing request interface. Details of the input / output processing request interface will be described later. When the disk input / output scheduler (104) starts next, the disk input / output processing request issued here is immediately processed. However, when issuing a disk input / output processing request, the disk input / output command list for the stream identifier indicated by the variable own stream identifier of the disk input / output command list header (501) must be empty. In the following description, it is assumed that the list is empty. If it is not empty, the reduction of response time, which is the object of the present invention, cannot be achieved. Step 606, Step 607 and Step 6
By issuing the disk I / O processing request according to the procedure of 09, the time until the disk I / O scheduler processes the request, that is, the response time to the issued disk I / O processing request is reduced.

At step 610, CPU handoff is performed.
This is to wait for completion of the disk input / output processing request issued in step 609.

At step 611, it is confirmed whether or not the disk input / output processing request issued at step 609 has been completed using the input / output processing completion interface. If not completed, CPU handoff is performed again (step 61).
2) Wait for completion of the disk input / output processing request issued in step 609. This I / O processing completion interface is only for inquiring of the disk I / O processing program whether the disk I / O processing request has been completed.

In step 613, it is determined whether or not all input / output for the variable's own stream identifier has been completed. If not, the process returns to step 606 to issue the next disk input / output processing request.

In step 614, the disk input / output module (103) is notified that the periodic input / output for the stream identifier indicated by the variable's own stream identifier has been completed using the periodic input / output completion interface.
At this time, the application program (102) also specifies a stream identifier as an argument. Details of the periodic input / output completion interface will be described later.

In step 615, the continuous media data is closed.

At step 616, the processing of the application program ends.

In the operation example shown in FIG. 6, the application program (102) is one piece of continuous media data (1
09), and periodically issues a disk input / output processing request. But the application program (102)
Can handle multiple continuous media data simultaneously. In this case, the application program may repeat steps 606 and 607 for all stream identifiers.

FIG. 7 shows an operation example of the periodic input / output request interface.

At step 702, an unused stream identifier is checked from the stream identifier management table (201). If not, step 706 returns NULL as the stream identifier to the caller of the interface. This means that there are no free time slots in the time slot table (107) to periodically process new continuous media data. Then, the processing of the periodic input / output request interface ends (step 707).

At step 703, the availability field (203) of the unused stream identifier in the stream identifier management table (201) is changed to "in use".

At step 704, a time slot whose stream identifier is X is found from the time slot table (107). Then, the stream identifier field (303) of the time slot is rewritten with the stream identifier assigned in step 703.

In step 705, the stream identifier assigned in step 703 is returned to the caller of the interface as a return value of the periodic input / output request interface.

In step 707, the processing of the periodic input / output request interface ends.

FIG. 8 shows an operation example of the periodic input / output completion interface.

In step 802, the stream identifier field (303) of the time slot corresponding to the stream identifier specified by the caller of the interface is changed to X in the time slot table (107).

In step 803, the usability field (203) corresponding to the stream identifier designated by the interface caller is changed to unused in the stream identifier management table (201).

In step 804, the processing of the periodic input / output completion interface ends.

FIG. 9 shows an operation example of the next stream reference interface (105).

In step 902, in the time slot table (107), the value of the stream identifier field (303) corresponding to the time slot indicated by the next time slot management variable (401) is recorded in the variable next stream identifier.

In step 903, the value of the variable next stream identifier is returned to the interface caller as a return value of the interface.

At step 904, the processing of the next stream reference interface (105) is completed.

FIG. 10 shows an operation example of the input / output processing request interface.

In step 1002, the disk input / output processing request specified by the argument by the caller of the interface is converted into a disk input / output command (504). If the disk I / O processing request cannot be processed in one time slot, the disk I / O processing request is divided into a plurality of disk I / O processing requests, and each is converted into a disk I / O command.

In step 1003, in the disk input / output command list header (501), the one created in step 1002 is added to the end of the disk input / output command list corresponding to the stream identifier specified by the argument by the caller of the interface. The above disk I / O commands are connected.

At step 1004, the processing of the input / output processing request interface ends.

FIG. 11 shows a disk input / output scheduler (1).
04) shows an operation example.

The disk input / output scheduler (104)
It is started periodically by the CPU scheduler (101). When the disk input / output scheduler (104) is activated for the first time, the process starts from step 1101. 2
The processing starts immediately after step 1108 from the first activation.

In step 1102, the stream identifier to be processed by the disk input / output scheduler (104) is obtained by using the next stream reference interface (105). It is recorded in the variable next stream identifier.

In step 1103, the disk input / output command list header (501) determines whether a disk input / output command to be processed exists in the disk input / output command list corresponding to the stream identifier indicated by the variable next stream identifier. Find out. If not, the process jumps to step 1105.

At step 1104, the first disk input / output command of the disk input / output command list corresponding to the stream identifier indicated by the variable next stream identifier is extracted from the disk input / output command list header (501). Then, the disk input / output command is passed to the disk input / output processing program (106). After the disk input / output scheduler (104) performs the CPU handoff, the disk input / output processing program (106)
Process the passed disk I / O command.

At step 1105, the value of the next time slot management variable (401) is increased by one.

At step 1106, it is determined whether or not the value of the next time slot management variable (401) is larger than the maximum value of the time slot in the time slot table (107). If it is larger, the value of the next time slot management variable (401) is returned to 1 in step 1107.

At step 1108, CPU handoff is performed.

With the above, the disk input / output scheduler (10
The processing in one cycle of 4) is completed. The disk input / output scheduler (104) itself resides in the computer system and is started periodically.

The following application example can be considered for the next stream reference interface (105) used in the above embodiment.

"Disk input / output processing start time reference interface" This is an interface for checking when a disk input / output processing request for a stream identifier specified by an argument is processed.

When the disk input / output processing start time reference interface is used, the application program (102) in FIG. 6 issues a disk input / output processing request when the disk input / output processing start time is approaching. The response time can be reduced in the same manner as in the embodiment.

In the above embodiment, all of the application programs (102) that periodically perform input / output
And the disk input / output scheduler (104) are activated in the same cycle. However, even when the periods are different from each other, if the operation example of the application program (102) in FIG. 6 is changed as follows,
The effect of shortening the response time can be obtained.

Instead of step 606, the following step 6
Step 17 is executed.

Step 617 Using the disk input / output processing start time reference interface, check the time at which the disk input / output processing request for the stream identifier to be input / output is processed.

Next, instead of step 607, step 6
Step 18 is executed.

Step 618 It is checked whether or not the time checked in step 617 is later than the time when the application program (102) is started next.

If the start time of the disk input / output processing is later, the application program (102) is activated at least once before the disk input / output processing request for the stream identifier is processed.
Therefore, step 608 is executed. If not, step 609 is performed. However, in order to realize step 618, the application program (10
The start time of 2) is changed by the application program (1).
02) needs an interface to know. This interface is easy to implement.

In the above embodiment, the values of the stream identifier fields (303) in the time slot table (107) do not overlap. However, step 70 in the periodic I / O request interface of FIG.
4 can be changed so that the stream identifiers are duplicated. In this case, the disk input / output processing request for the stream identifier is processed only for the duplication.
This corresponds to handling various playback speeds in moving images, for example. Also in this case, the response time can be reduced by the above-described embodiment. However, at the time of assigning an unused stream identifier in step 703 of FIG. 7, it must be determined whether a new periodic input / output request can be accepted. However, before that, the application program (102) specifies the playback speed of the continuous media data as an argument when calling the periodic input / output request interface. The algorithm for determining whether or not it can be accepted can be considered in the same way as the CPU scheduling problem.

As another application example of the present invention, it is possible to consider a case where the number of disk devices (108) is plural and the continuous media data (109) is stored in stripes on a plurality of disk devices. . Again, time
The slot tables (107) are prepared for the number of disk devices, and the next time slot management variables (40
What is necessary is just to provide 1). In the next stream reference interface (105), the application program (102) refers to the data for which a disk input / output processing request is to be issued as an argument, and checks the time slot table (107) for the disk in which it exists. Unlike FIG. 6, the application program (10
2) Before executing the next stream reference interface (105), determine the data for which a disk input / output processing request is to be issued. Other than that, it is the same as the above embodiment.

[0096]

According to the present invention, the timing at which an application program issues a disk input / output processing request,
By synchronizing the timing at which the disk input / output scheduler processes the disk input / output processing request, the response time of the disk input / output processing can be reduced.

[Brief description of the drawings]

FIG. 1 is a diagram showing a processing flow of a disk input / output processing request for realizing a disk input / output method.

FIG. 2 is a diagram showing an example of a stream identifier management table.

FIG. 3 is a diagram showing an example of a time slot table.

FIG. 4 is a diagram showing next time slot management variables.

FIG. 5 is a diagram showing an example of a disk input / output command list header.

FIG. 6 is a flowchart of an application program.

FIG. 7 is a flowchart of a periodic input / output request interface.

FIG. 8 is a flowchart of a periodic input / output completion interface.

FIG. 9 is a flowchart of a next stream reference interface.

FIG. 10 is a flowchart of an input / output processing request interface.

FIG. 11 is a flowchart of a disk input / output scheduler.

FIG. 12 is a diagram showing a hardware configuration.

FIG. 13 is a diagram showing a relationship between a scheduling table and a time slot table.

FIG. 14 is a diagram showing a conventional video library system.

FIG. 15 is a diagram showing a timing chart of disk input / output.

FIG. 16 is a diagram showing a timing chart of disk input / output with reduced response time.

FIG. 17 is a diagram showing an example in which the scheduling table and the time slot table are not synchronized.

FIG. 18 is a diagram showing an example in which a scheduling table and a time slot table are synchronized.

[Explanation of symbols]

101: CPU scheduler, 102: application program, 104: disk input / output scheduler,
Reference numeral 105: next stream reference interface, 107: disk I / O scheduling table, 108: disk, 109: stream file, 201: stream identifier management table, 401: next time slot management variable, 501: disk I / O command list -Header, 504: Disk input / output command, 508 ...
Disk I / O command list.

Continuation of the front page (72) Inventor Takahiro Nakano 1099 Ozenji Temple, Aso-ku, Kawasaki City, Kanagawa Prefecture F-term in Hitachi, Ltd. System Development Laboratory 5B065 BA01 CE01 CH13 CH14 5B098 AA05 GA01 GA02 GB01 GB11 GC05 GD03 GD07 GD12 GD14 5C064 BA01 BB05 BC18 BD13 5D044 AB05 AB07 BC01 BC04 CC04 FG21 HL11

Claims (1)

[Claims] 1. A plurality of continuous media data is recorded on a plurality of disk devices, and a time time indicating when and which continuous media data is to be subjected to disk input / output processing.
In a computer system having a slot table and processing a disk I / O processing request based on the time slot table, an application program specifies an identifier of continuous media data, and a disk input for the continuous media data is specified. A processing unit for calculating a start time of the output process; a means for referring to the start time by the application program; a processing unit for calculating a time at which the application program is next started; A means for referring to the time by the application program; and a processing unit for comparing the start time with the start time. If the start time is earlier, the application program inserts the disk into the continuous media data. Having means for issuing an output processing request. Disk input and output method.