JP2003037649A - Method for estimating contents distribution end time, recording medium and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To predict an effective throughput for the unit of a flow while considering the influence of a TCP(transmission control protocol), to provide sufficient measurement accuracy on various access lines and further to suppress increase of influence upon a network and time required for measurement. SOLUTION: In the method for predicting the contents distribution end time to be used for estimating the distribution end time of the contents when transferring the information of digital contents from a transmitting station through the Internet to a receiving station, a test message of the length of a predetermined first size is transferred from the transmitting station to the receiving station in a preliminary test, the effective throughput is detected in a primary measured value on the basis of time required for test message transfer in the preliminary test, a second size is determined on the basis of the effective throughput of the primary measured value after the preliminary test, a test message of the length of the second size is transferred from the transmitting station to the receiving station in the main test, and an effective throughput is detected in a secondary measured value on the basis of time required for test message transfer in the main test.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a content distribution completion time estimation method used for estimating a content distribution completion time when transferring digital content information from a transmitting station to a receiving station via the Internet. And a recording medium and a program.

[0002]

2. Description of the Related Art For example, a user is a user terminal (transmitting station).
In order to publish the above digital contents, information of the digital contents may be transferred (uploaded) from a user terminal to a file server (reception station) on the Internet. Further, when the amount of information of the content to be transferred is relatively large, it takes time to transfer the data, so it is desirable to be able to grasp the scheduled delivery completion time of the content before starting the transfer.

However, transmission resources are not allocated to each communication session on the Internet, and communication quality such as communication delay and throughput is not guaranteed. Therefore, the communication quality changes every moment depending on the load condition of the network. Therefore, the time required to transfer the content data on the Internet is
It changes depending not only on the data size of the content but also on the effective throughput which changes from moment to moment. for that reason,
In order to grasp the scheduled delivery completion time of the content before starting the transfer, it is necessary to predict the effective throughput by some method before starting the transfer.

There are several methods for examining the effective throughput, but they can be roughly classified into a passive type that does not generate new traffic during measurement and an active type that does generate new traffic. The passive type has the following methods (1) and (2), and the active type has the following methods (3) and (4).

(1) The effective throughput in the past content transfer is calculated by analyzing the log of the WWW server. (2) A probe is attached to a node, a session of content transfer such as HTTP is identified from the header information of a packet flowing on the line, and the effective throughput is calculated based on the transfer time and the transfer file size.

(3) "netperf", "tren"
TCP / I using a tool (program) such as "o"
The message transfer by P is actually performed and the effective throughput is measured. (4) An ICMP packet is sent to each node that passes through when the content is transferred, the waiting time at each node is measured from the delay time, and the effective throughput of each routed link is derived.

[0007]

However, the conventional methods shown in the above (1) to (4) have the following problems, respectively. (1) Although the effective throughput between end hosts including the influence of TCP (Transmission Control Protocol) flow control can be obtained, the measurement target is limited to only past flows remaining in the log of the WWW server.

(2) Although the effective throughput between end hosts including the effect of TCP flow control can be obtained, an extra cost is required to install the probe. Moreover, the measurement target is limited to the past flow. (3) The effective throughput between arbitrary hosts including the influence of TCP flow control can be obtained most directly. However, it is necessary to generate a large amount of extra traffic for measurement in the network. If the test message is made too small to suppress the traffic volume, the measurement accuracy is likely to deteriorate.

(4) It is possible to reduce the amount of extra traffic generated by the measurement. However, it is not possible to obtain the effective throughput of each flow unit including the influence of TCP flow control. Since TCP / IP, which is a standard protocol of the Internet, is used for transferring contents, T is used when estimating the effective throughput.
It is necessary to consider the impact of CP flow control.

What is common to the passive type measurement methods is that the network does not generate extra traffic for measurement, but the measurement accuracy is limited because the measurement target is limited to the past flow. There is a problem of doing. The present invention, in the method of estimating the content delivery completion time, predicts the effective throughput of each flow in consideration of the influence of TCP, realizes sufficient measurement accuracy for various access lines, and has an effect on the network and The purpose is to suppress an increase in the time required for measurement.

[0011]

According to a first aspect of the present invention, a content distribution completion time used for estimating a content distribution completion time when transferring digital content information from a transmitting station to a receiving station via the Internet. As a preliminary test, a test message of a predetermined first size length is transferred from the transmitting station to the receiving station, and is effective based on the time required for transferring the test message in the preliminary test. Throughput is detected as a primary measurement value, and after the preliminary test, the second size is determined based on the effective throughput of the primary measurement value.
The test message having the length of the second size is transferred from the transmitting station to the receiving station, and the effective throughput is detected as a secondary measurement value based on the time required for transferring the test message in the main test. .

When a relatively small test message is transmitted in order to measure the throughput, the transfer is completed and the measurement result is obtained before the congestion window used for TCP flow control is sufficiently widened, so that it is detected. The throughput is lower than when actually transferring the information of the digital contents. That is, the throughput measurement accuracy is poor, and the content distribution completion time cannot be correctly estimated.

Further, as the effective throughput in the direction from the transmitting station to the receiving station increases, the time required to complete the expansion of the congestion window becomes longer. Therefore, when the actual effective throughput is large, the effective throughput cannot be measured with sufficient accuracy unless a large test message is transferred.

However, when a large test message is transferred via a communication path having a small effective throughput, the time required for measurement is significantly increased. Therefore,
It is difficult to achieve both sufficient measurement accuracy and suppression of required measurement time. Therefore, in claim 1, prior to the main test of the throughput measurement, a preliminary test is performed for the purpose of roughly identifying the effective throughput. In this preliminary test, a test message of a predetermined first size length is used. Since high measurement accuracy is not required in the preliminary test, it is desirable to use a relatively small size test message to reduce the measurement time.

Since the condition of the communication path (approximate effective throughput) can be grasped from the result of the preliminary test, an appropriate length (second size) of the test message in the main test can be determined based on it. For example, if the effective throughput is relatively large, considering that it takes a long time to complete the expansion of the congestion window,
Large test messages can be transferred in this test. Further, when the effective throughput is relatively small, a small test message may be transferred in this test in consideration of the short time required to complete the expansion of the congestion window.

Therefore, it is possible to transfer the test message of the minimum required size in the main test, and to achieve both sufficient measurement accuracy and suppression of the required measurement time. The content distribution completion time can be obtained based on the size of the information of the digital content and the detected effective throughput. According to a second aspect of the present invention, in the content distribution completion time estimation method according to the first aspect, a plurality of size candidates corresponding to the second size are previously determined, and an effective throughput of the primary measurement value is at least predetermined. One of the plurality of size candidates is compared with one threshold value according to the comparison result.
One is selected as the second size.

According to the second aspect, the second size can be easily selected from a plurality of size candidates prepared in advance, based on the result of comparison between the effective throughput of the primary measurement value and the threshold value. According to a third aspect of the present invention, in the content distribution completion time estimation method according to the first aspect, the effective throughput of the primary measurement value is compared with a predetermined threshold value, and when the effective throughput is equal to or less than a predetermined threshold, the book The test is omitted, and the primary measurement value is used as the secondary measurement value.

When the effective throughput is relatively small, the time required for the expansion of the congestion window to be completed is short, and therefore a highly accurate effective throughput can be obtained by simply transferring a relatively small test message. In such a case, the accuracy of the effective throughput itself of the primary measurement value obtained in the preliminary test is expected to be high, and therefore there is no problem even if this test is omitted.

According to the third aspect, since the present test is omitted when the effective throughput is relatively small, the time required for estimating the content distribution completion time can be shortened. A fourth aspect of the present invention is a recording medium recording a computer-readable program used for estimating a distribution completion time of content when transferring digital content information from a transmitting station to a receiving station via the Internet. As a preliminary test, the program is based on a procedure of transferring a test message of a predetermined first length from the transmitting station to the receiving station and a time required for transferring the test message in the preliminary test. Detecting the effective throughput as a primary measurement value, determining the second size based on the effective throughput of the primary measurement value after the preliminary test, and as the main test, the length of the second size. Based on the procedure for transferring the test message from the transmitting station to the receiving station and the time required for transferring the test message in the main test described above. Characterized by providing a procedure for detecting the throughput as a secondary measure.

By reading the program recorded in the recording medium of claim 4 and executing the program by the computer, the method of estimating the content distribution completion time according to claim 1 can be implemented. According to a fifth aspect of the present invention, in the recording medium according to the fourth aspect, the program has at least predetermined data of a plurality of predetermined size candidates corresponding to the second size and an effective throughput of the primary measurement value. A procedure of comparing with one threshold value and selecting one of the plurality of size candidates as the second size according to the comparison result is further provided.

By reading the program recorded in the recording medium of claim 5 and executing the program by the computer, the method of estimating the content distribution completion time according to claim 2 can be implemented. According to a sixth aspect of the present invention, in the recording medium of the fourth aspect, the program compares the effective throughput of the primary measurement value with a predetermined threshold value, and when the effective throughput is less than or equal to a predetermined value, the main test is performed. Is omitted, and a procedure for using the primary measurement value as a secondary measurement value is further provided.

By reading the program recorded in the recording medium of claim 6 and executing it by the computer, the method of estimating the content distribution completion time of claim 3 can be implemented. A seventh aspect of the present invention is a computer-readable program used for estimating a distribution completion time of content when transferring digital content information from a transmitting station to a receiving station via the Internet. , Detecting the effective throughput as a primary measurement value based on the procedure of transferring a test message of a predetermined first length from the transmitting station to the receiving station and the time required for transferring the test message in the preliminary test. After the preliminary test, the second size is determined based on the effective throughput of the primary measurement value after the preliminary test, and as a main test, a test message having the length of the second size is transmitted from the transmitting station to the receiving station. The effective throughput is detected as the secondary measurement value based on the transfer procedure and the time required to transfer the test message in the main test. Characterized by providing a procedure.

By loading the program according to claim 7 and executing it on a computer, the method for estimating the content distribution completion time according to claim 1 can be implemented. Claim 8 compares the data of a plurality of predetermined size candidates corresponding to the second size with the program of claim 7, and compares the effective throughput of the primary measurement value with at least one predetermined threshold value. And a procedure for selecting one of the plurality of size candidates as the second size according to the comparison result.

By loading the program according to claim 8 and executing the program on a computer, the method for estimating the content distribution completion time according to claim 2 can be implemented. According to a ninth aspect of the present invention, in the program of the seventh aspect, the effective throughput of the primary measurement value is compared with a predetermined threshold value, and when the effective throughput is equal to or less than a predetermined threshold,
The procedure of omitting the execution of the main test and using the primary measurement value as the secondary measurement value is further provided.

By loading the program of claim 9 and executing it on a computer, the method of estimating the content distribution completion time of claim 3 can be implemented.

[0026]

BEST MODE FOR CARRYING OUT THE INVENTION One embodiment of a method for estimating a content distribution completion time, a recording medium and a program according to the present invention will be described with reference to FIGS.
This form corresponds to all the claims.

The present invention is carried out in a system having a structure as shown in FIG. 4, for example. In the example of FIG. 4, the user terminal 11 and the file server 12 are connected to the Internet 13. In this example, the file server 12 stores home page information of various users. The file server 12 provides the accumulated information to various users who access it via the Internet 13.

New information (digital contents) is stored in the file server 12 regarding the home page of a specific user.
When uploading, the user having a specific ID or password transfers the digital content data from the user terminal 11 to the file server 12, for example. This digital content is transferred from the user terminal 11 to the file server 12 via the Internet 13. The time required for this transfer greatly changes depending on the amount of information of digital contents and the state of the communication line used (effective throughput: transfer rate).

That is, when transferring a digital content having a very large amount of information or when the effective throughput is slow, the time required from the start of the transfer of the digital content to the completion of the transfer becomes very long. Therefore, it is desirable for the user to know the content distribution completion time before starting the transfer of the digital content. The content distribution completion time can be predicted based on the amount of information and the effective throughput of the digital content to be transferred. However, since the effective throughput changes sequentially, it is necessary to actually measure the effective throughput before transferring digital contents.

In this embodiment, a recording medium (for example, CD-ROM) in which a program for estimating the content distribution completion time is recorded is prepared in advance, and the program is read and executed by the computer of the user terminal 11. By executing this program, the user of the user terminal 11 can grasp the content distribution completion time before transferring the digital content.

This program executes the processing shown in FIG. 1 in order to measure the effective throughput. In this process, a test message is sent from the user terminal 11 which is the transmitting station to the file server 12 which is the receiving station by TCP
(Transmission Control Protocol)
Measure the transfer time.

The test message used for transfer is composed of dummy data (information of meaningless character string) having a specific message size as shown in FIG. A TCP header is added to this test message to form a TCP segment. In addition, TCP segment to IP (Internet Proto
col) packet is generated as shown in FIG.
The packet is transferred on the Internet 13.

In order to measure the effective throughput, it is necessary for the transmitting station and the receiving station to recognize that the test message is the data for the test. Therefore, a specific port number for measuring the effective throughput is prepared, and the port number is specified in the TCP header of the test message. As shown in FIG. 1, a preliminary test is first performed to measure the effective throughput, and a main test is performed based on the result. In the preliminary test, a relatively small size test message is transferred, and in this test, the size of the test message is determined according to the result of the preliminary test. Further, when the effective throughput is small, the effective throughput can be measured with high accuracy only by the preliminary test. In that case, the main test is omitted.

In step S11 of FIG. 1, a test message having a predetermined message size Xp is acquired. For example, in the example of FIG. 5, a test message of 80 kbytes is used. In step S12, transfer of the test message of size Xp acquired in step S11 is started. This test message is transferred from the transmitting station (11) to the receiving station (12). At the same time when the transfer is started, step S1
At 3, the timer for time measurement is started.

As shown in FIG. 3, when the transfer of the test message is completed, the receiving station transmits a reception response to the transmitting station. At the transmitting station that executes the processing of FIG.
When a response from the receiving station is detected in step 4, the next step S1
The process proceeds to step 5 to detect the transfer time Tp. That is, as shown in FIG. 3, the transfer time Tp from the start of the transfer of the test message of the preliminary test to the reception of the transfer completion response is read from the measured value of the timer started in step S13.

The effective throughput Bp in the preliminary test is obtained by dividing the size Xp of the transferred test message by the transfer time Tp. However, since the preliminary test uses a test message having a relatively small size Xp, the effective throughput Bp obtained in the preliminary test is not very accurate especially when the actual effective throughput is large. However, it takes a long time to measure when a large test message is transferred.

Therefore, a main test is carried out in order to measure the accurate effective throughput. The size of the test message used in this test is the effective throughput Bp obtained in the preliminary test.
According to. In this example, the effective throughput Bp
To partition the size of into any of the N regions,
(N-1) thresholds B (1) to B (N-1) are prepared.

In steps S16, S17 and S18 of FIG. 1, the effective throughput Bp is compared with the threshold values B (1) to B (N-1). In addition, in FIG. 1, description of a part of the processing is omitted. For example, in the example of FIG. 5, two thresholds of 64 kbps
The effective throughput Bp is divided into one of the three areas using s and 128 kbps.

In FIG. 1, when "Tp <B (1)", the process directly proceeds from step S16 to step S26, and thus the main test is omitted. If “B (2) ≦ Tp <B (1)”, the process proceeds from step S16 to step S17 and then to step S19, so that the test message of message size X (2) is acquired. The message size X (2) is “16 in the example of FIG.
0 kbyte ".

Similarly, in the case of "B (N-1) ≤Tp <B (N-2)", steps S16, S17, S18 and S20 are executed.
Then, the test message of the message size X (N-1) is acquired. If "B (N-1) ≤Tp", the process proceeds to S21 through steps S16, S17, S18, and the test message of message size X (N) is acquired. The message size X (N) is “3” in the example of FIG.
20 kbytes ".

In step S22, each step S19,
The transfer of the test messages of different sizes acquired in S20 and S21 is started. This test message is transferred from the transmitting station (11) to the receiving station (12).
At the same time when the transfer is started, a timer for measuring time is started in step S23. As shown in FIG. 3, also in this test, when the transfer of the test message is completed, the receiving station transmits a reception response to the transmitting station.

In the transmitting station which executes the processing of FIG. 1, when the response from the receiving station is detected in step S24, the process proceeds to the next step S25 and the transfer time Tm is detected. That is, as shown in FIG. 3, the transfer time Tm from the start of transfer of the test message of this test to the reception of a transfer completion response.
Is read from the measured value of the timer started in step S23.

In step S26, the effective throughput is calculated based on the transfer time Tm of the main test obtained in step S25. That is, the effective throughput Bm in this test can be obtained by dividing the size “X (2) to X (N)” of the test message used in this test by the transfer time Tm.
However, if “Tp <B (1)” in step S16, the effective throughput Bp obtained in the preliminary test is directly used as the effective throughput Bm in the main test.

Since the time required for delivery of the digital content to be transferred by the user is obtained by dividing the information amount of the digital content by the effective throughput Bm, the content delivery completion time can be estimated from the required time. The size Xp of the test message used in the preliminary test, the optimum value of each size X (2) to X (N) of the test message used in this test, the number N of effective throughput categories and the thresholds B (1) to B (N) The optimum value of -1) is affected by various factors such as network topology, traffic pattern, physical link rate, routing protocol, router processing capacity, etc.

Therefore, regarding the size of each test message, the number of effective throughput divisions, and the threshold value, the effective throughput measurement test is performed in advance for some states of the network to which the present invention is actually applied, and the measurement accuracy and measurement Both sides need to decide to meet the demand. In the above embodiment, the size of the test message of the main test is fixedly determined according to the result of comparison between the effective throughput Bp obtained in the preliminary test and the threshold value. It is also possible to calculate the size of the test message of this test from the effective throughput Bp using.

It is needless to say that the method of estimating the content delivery completion time according to the present invention can be used for various data transfer applications other than the case where the user uploads the digital content to the file server as in the embodiment. Yes.

[0047]

As described above, according to the present invention, by transferring the test message, the effective throughput in consideration of the influence of TCP can be measured. Moreover, since the test message is transferred in two stages of the preliminary test and the main test, the test message of the minimum required size can be used in the main test, and the effective throughput can be measured with high accuracy in a relatively short measurement time. . The influence of the test message on the network is also suppressed.

[Brief description of drawings]

FIG. 1 is a flowchart showing an effective throughput measurement process.

FIG. 2 is a schematic diagram showing the structure of a test message.

FIG. 3 is a sequence diagram showing a transfer time measurement sequence of a test message.

FIG. 4 is a block diagram showing a configuration example of a system to be used.

FIG. 5 is a schematic diagram showing a specific example of effective throughput measurement conditions.

[Explanation of symbols]

11 user terminal 12 file server 13 Internet

   ─────────────────────────────────────────────────── ─── Continued front page    F term (reference) 5K030 GA14 HA08 HB06 HC01 JA10                       MB06 MC03                 5K035 AA03 BB02 EE00 GG03 HH07

Claims (9)

[Claims] 1. When transferring digital content information from a transmitting station to a receiving station via the Internet,
A method for estimating a content delivery completion time used for estimating a content delivery completion time, wherein as a preliminary test, a test message of a predetermined first length is transferred from a transmitting station to a receiving station, and The effective throughput is detected as the primary measurement value based on the time required for the test message transfer in the preliminary test, and the second size is determined based on the effective throughput of the primary measurement value after the preliminary test, and the main test is performed. As a result, the test message of the second size length is transferred from the transmitting station to the receiving station, and the effective throughput is detected as a secondary measurement value based on the time required for transferring the test message in the main test. And the method of estimating the content delivery completion time. 2. The content distribution completion time estimation method according to claim 1, wherein a plurality of size candidates corresponding to the second size are previously determined, and an effective throughput of the primary measurement value is at least predetermined. A method of estimating a content distribution completion time, which comprises comparing one threshold value and selecting one of the plurality of size candidates as the second size according to a comparison result. 3. The content delivery completion time estimation method according to claim 1, wherein the effective throughput of the primary measurement value is compared with a predetermined threshold value, and when the effective throughput is equal to or less than a predetermined threshold, A method of estimating a content delivery completion time, characterized in that the test is omitted and the primary measurement value is used as a secondary measurement value. 4. When transferring digital content information from a transmitting station to a receiving station via the Internet,
A recording medium recording a computer-readable program used for estimating a distribution completion time of content, wherein the program transmits a test message of a predetermined first length as a preliminary test. From the station to the receiving station, the procedure of detecting the effective throughput as the primary measurement value based on the time required for the test message transfer in the preliminary test, and the effective of the primary measurement value after the preliminary test. The procedure for determining the second size based on the throughput, the procedure for transferring the test message of the length of the second size from the transmitting station to the receiving station as the main test, and the test message transfer in the main test were required. And a procedure for detecting an effective throughput as a secondary measurement value based on time. 5. The recording medium according to claim 4, wherein the program includes at least data of a plurality of predetermined size candidates corresponding to the second size and an effective throughput of the primary measurement value. A recording medium further comprising a step of comparing with one threshold value and selecting one of the plurality of size candidates as the second size according to a comparison result. 6. The recording medium according to claim 4, wherein the program compares the effective throughput of the primary measurement value with a predetermined threshold value, and when the effective throughput is less than or equal to a predetermined value, the main test is performed. The recording medium characterized by further comprising the step of omitting the step of (1) and using the primary measurement value as a secondary measurement value. 7. When transferring digital content information from a transmitting station to a receiving station via the Internet,
It is a computer-readable program used for estimating the delivery completion time of content, and as a preliminary test, a procedure of transferring a test message of a predetermined first size length from the transmitting station to the receiving station, A procedure of detecting an effective throughput as a primary measurement value based on the time required for transferring a test message in the preliminary test, and determining a second size based on the effective throughput of the primary measurement value after the preliminary test. As a procedure, as a main test, the effective throughput is secondarily measured based on the procedure for transferring the test message of the second size from the transmitting station to the receiving station and the time required for transferring the test message in the main test. A program which is provided with a procedure of detecting as a value. 8. The program according to claim 7, wherein data of a plurality of predetermined size candidates corresponding to the second size and the effective throughput of the primary measurement value are compared with at least one predetermined threshold value. And a procedure for selecting one of the plurality of size candidates as the second size according to a comparison result. 9. The program according to claim 7, wherein the effective throughput of the primary measurement value is compared with a predetermined threshold value, and when the effective throughput is equal to or less than a predetermined value, the implementation of the main test is omitted, A program further comprising a procedure of using the primary measurement value as a secondary measurement value.