JPH10340236A - Method for detecting host short in access time of http communication - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make connectable with an easy-to-connect host and to make Internet-connecting operation smooth, by monitoring the state wherein an HTTP client accesses an HTTP server and measuring the time of access to each server, and detecting a host, which is short in access time from the measured value. SOLUTION: A measurement server 3 provided between an HTTP client 1 and an HTTP server 5 records the communication destination server names, request times, and access time of all communications, which can be monitored between the HTTP client 1 and HTTP server 5. Here, a measurement server 3 monitors the state wherein the HTTP client 1 as a user terminal accesses the HTTP server 5 and measures the time of access to each server in an HTTP communication. Then, the reciprocation time of a packet is calculated from the measured value of the access time to detect a host, which is short in the time of access from the network of the user.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to an HTTP (Hypert
ext Transport Protocol) HTTP communication based on the round trip time of the packet measured from the HTTP client side.
The present invention relates to a method for detecting a host having a short access time in communication.

[0002]

2. Description of the Related Art At present, there are many WWs on the Internet.
A W (World Wide Web) server is installed and used as a main service of the Internet. This WWW
Access to the server is realized by HTTP.

[0003] With the increase of multimedia contents such as WWW, access to the Internet is increasing. In particular, in the case of communication using HTTP, which is a highly interactive communication method, the waiting time at the time of access increases, which causes user stress. That is, the waiting time at the time of access is an important index for judging the comfort of the network for the user.

[0004]

At present, in order to check the communication state between hosts, which are an unspecified number of WWW servers on the Internet, measurement of response time using ICMP,
There is only a method to actually access like SNMP network method etc. and guess from the result,
Such a method is not only difficult to investigate because it involves the traffic for the investigation, but also may cause a load on the entire network, and it is difficult to apply in practice. There's a problem.

[0005] The present invention has been made in view of the above,
The purpose is to use HTTP communication in HTTP communication.
H measures the round trip time of the packet from the P client side and detects a host with a short access time from this round trip time.
An object of the present invention is to provide a method for detecting a host having a short access time in TTP communication.

[0006]

In order to achieve the above-mentioned object, the present invention according to claim 1 is characterized in that a user terminal (HTT) is provided.
The P client monitors the situation of accessing the HTTP server which is a WWW server, measures the access time to each server in the HTTP communication, calculates the packet round trip time from the measured value of the access time, The gist is to detect a host having a short access time.

[0007] In the first aspect of the present invention, the HTT
Automatically connect to a host that is easily connected to monitor the situation where the P client accesses the HTTP server, measure the access time to each server, and detect the host with the short access time from the measured access time. It is no longer possible to operate smoothly even when connected to the Internet as in the past.

[0008]

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

FIGS. 1A and 1B are diagrams for explaining a method of detecting a host having a short access time in HTTP communication according to an embodiment of the present invention. In FIG. 1A, a plurality of HTTP clients 1 are connected to an HTTP server 5 via a measurement server 3, whereby H
The measurement server 3 monitors the access status from the plurality of HTTP clients 1 to the HTTP server 5 in the HTTP communication, and measures the access time in the HTTP communication.

That is, when a user accesses a page of a WWW server from a certain terminal, the user terminal uses an HTT
It is a P client and the WWW server is an HTTP server. The access time in the communication between the HTTP client and the HTTP server depends on the user terminal HT.
When viewed from the TP client side, as shown in FIG. 1B, the HTTP client 1 transmits an HTTP request to the HTTP server 5, and the first HTTP from the HTTP server 5 that has received the HTTP request.
"Request time (t
1) "and" Data transfer time (t) indicated by hatching from the reception of the first response to the end of all communications.
1-t2) ".

In the HTTP communication, the required data size differs for each communication, so that the data transfer time is affected by the size of the target data. On the other hand, the HTTP request transmitted from the HTTP client 1 is about 300 to 500 bytes regardless of the size of the target data, and the communication ends in almost one packet.

Therefore, in FIG. 1, the measurement server 3 provided between the HTTP client 1 and the HTTP server 5 performs the communication destination server name for all the communications that can be monitored between the HTTP client 1 and the HTTP server 5. , Request time t1, and access time. Since the measurement server 3 is installed on the same network as the HTTP client 1, the request time measured by the HTTP client 1 and the request time measured by the measurement server 3 are almost equal.

Next, with reference to the flowchart shown in FIG. 2, a description will be given of the process of summing up the request time data for each communication destination server.

The communication results measured as described above are divided into certain target time periods, such as 60 minutes or 30 minutes, for example, from 4:00 pm to 5:00 pm, and access records are extracted and counted (step S11), and communication is performed. The number of all communications made to the destination is recorded, and this is set as Z (step S13). Then, the request time s is set to 1 second (step S1).
5) The number of communications having a request time s (= 1) seconds is calculated and recorded as Y (step S17).
Then, the ratio Y / Z at which the request time s thus calculated is present is calculated (step S19), and then the request time s is incremented by +1 (step S2).
3) Return to step S17, and calculate the ratio of the number of communications in each request time s while sequentially increasing the request time.

That is, from the communication results totaled by the measurement server 3, all the addresses addressed to the same communication destination server are extracted, and the number of communication for each request time, such as a request time of 1 second or a request time of 2 seconds, is calculated. Tally. Then, what percentage of the number of seconds in each request time is the percentage of the communication performed in the time zone is calculated.

FIG. 3 is a table showing an example of statistics in which the ratio of the number of communications is counted for each request time as described above. In this table, when the request time of the servers A, B, and C is 1 second, the ratio of the number of communications is 81%, 91%, respectively.
Server C is the most crowded,
It can be seen that the access time to the server B is the shortest.

Next, with reference to a flowchart shown in FIG. 4, a process of sorting hosts in shortest request time in order to detect a host having a short access time, that is, an HTTP server will be described.

As described above, when the process of totalizing the number of communication and the ratio thereof is completed, the destination host name, that is, the destination server name is sorted by the percentage of the request time of 1 second (step S31), Is detected as the server having the shortest access time, the server having the largest percentage, which is the ratio in the case where is 1 second (step S3).
3). If the percentages when the request time is 1 second are equal (step S35), the server with the next largest percentage when the request time is 2 seconds is detected as the server with the shortest access time (step S3).
7). The above processing is performed for each list subjected to the tallying processing (step S39).

That is, the server having the largest ratio when the request time is 1 second is selected. When the ratio when the request time is 1 second is the same, the ratio when the request time is 2 seconds is selected. By selecting the larger one, it is possible to know the server with the shorter access time among the other servers of the communication performed in the time zone.
Further, when communication with the same host is frequently performed, the transition of the request time according to the time zone can be obtained.

Next, referring to FIGS. 5 and 6, WWW
Access when the server has a mirror server will be described.

Most of the popular WWW servers with a high number of accesses now have a mirror server. This aims at distributing server load by distributing accesses by installing multiple servers having the same information.

5 and 6, the WWW server X is provided with three mirror servers A, B and C. In such a state, as shown in FIG. 5, when the access from the HTTP client is not to be sorted, the HTTP
The client will access the WWW server X.

On the other hand, as shown in FIG. 6, when the measurement server 3 is provided between the HTTP client and the WWW server X, the measurement server 3 includes the mirror servers A, B, and C. If information about the server which is a host having a short access time of the WWW server X where the server is installed is stored, and a function of controlling access through HTTP communication is further provided, as will be described below,
For access from the TP client, a mirror server having a short access time is selected from a plurality of mirror servers A, B, and C.
It is sorted to perform communication to the server with the short access time.

That is, if the HTTP client is WWW
When an HTTP request is transmitted to the server X, the measurement server 3 shown in FIG. 6 confirms that the WWW server X is a host having a mirror server, temporarily stops access from the HTTP client, and stops measurement. Looking at the information of the mirror servers A, B, and C of the WWW server X stored in the server 3, the mirror server having the shortest access time is checked, and it is found that the mirror server B has the shortest access time.

The measurement server 3 has been the WWW server X
The HTTP request transmitted to the mirror server B is transmitted to the mirror server B. As a result, the HTTP client can communicate with the mirror server B having the shortest access time. In this way, the user can reduce the access time without any special operation or consciousness.

[0026]

As described above, according to the present invention,
It monitors the situation in which the HTTP client accesses the HTTP server, measures the access time to each server, and detects the host with the short access time from the measured value of the access time, so that it automatically connects to a host that is easily connected. This makes it possible to prevent the user from moving smoothly even when connected to the Internet as in the related art, and is useful for selecting a connection destination, selecting a connection time zone, and the like.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining a method for detecting a host having a short access time in HTTP communication according to an embodiment of the present invention.

FIG. 2 is a flowchart illustrating a process of summing up request time data for each communication destination server.

FIG. 3 is a table showing an example of statistics in which the ratio of the number of communications is counted for each request time.

FIG. 4 is a flowchart illustrating a process of sorting hosts in ascending request time in order to detect a host having a short access time.

FIG. 5 is a diagram showing access from an HTTP client in a case where distribution is not performed in a WWW server having a mirror server.

FIG. 6 is a diagram illustrating access from an HTTP client when distribution is performed in a WWW server having a mirror server.

[Explanation of symbols]

 1 HTTP client 3 Measurement server 5 HTTP server

Claims (1)

[Claims] An HTTP client serving as a user terminal monitors a situation where an HTTP server serving as a WWW server is accessed, an access time to each server in HTTP communication is measured, and a packet round-trip time is calculated from the measured access time. And detecting a host having a short access time in the HTTP communication, wherein the host detects a host having a short access time as viewed from the own network.