JP2017182157A - Web content distribution method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To speed up Web content distribution capable of extending an initial congestion window size of a single TCP connection in a pseudo manner by establishing a plurality of SPDY connections between a client and a server.SOLUTION: A client 1 establishes a preceding SPDY session with a server 2 to request a configuration file of content [Fig.1(a)]. The client 1 executes a procedure for additionally establishing a plurality of following SPDY sessions with the server 2 during a response delay period since the establishment of the preceding SPDY session is requested until the configuration file is received [Fig.1(b)]. When the plurality of following SPDY sessions are established, the client 1 which has received the configuration file uses the preceding SPDY session which has finished transmitting and receiving the configuration file and the plurality of following SPDY sessions to request and receive a plurality of sub-content described in the configuration file in parallel [Fig.1(c)]SELECTED DRAWING: Figure 1

Description

  The present invention relates to a Web content distribution method for a client to receive Web content from a server at high speed, and in particular, Web content that enables high-speed Web content distribution by artificially expanding the initial congestion window size. Regarding the delivery method.

  HTTP version 1.1, which is widely used as a web content distribution protocol in the current Internet, speeds up the display of a web page including a large number of image contents mainly by a keep-alive function. In HTTP / 1.1, the number of connections that can be connected simultaneously was small at the beginning, and a multi-domain web page was developed as a technology to compensate for this. However, with the keep-alive function, after the request is transmitted, the next request cannot be transmitted until the response is received.

  In order to solve such technical problems, HTTP / 2.0 has been studied as a next generation version of HTTP. This HTTP / 2.0 solves the above technical problem by establishing a plurality of HTTP sessions in parallel on a single TCP connection.

  In Non-Patent Document 1, by multiplexing HTTP sessions over a single TCP connection, prioritizing HTTP requests, omitting redundant headers, making communication data compression mandatory, and push delivery from the server, A technique for improving the Web content delivery speed in the existing HTTP is disclosed. This technology is implemented in the session layer, thereby realizing the functions described above without modifying the existing TCP, SSL, and HTTP.

  Non-Patent Document 2 discloses a technology that applies MPEG-DASH (Dynamic adaptive streaming over HTTP) that performs streaming distribution using HTTP to SPDY distribution. In MPEG-DASH, the content is held (in the server) at a plurality of encoding rates according to the network bandwidth for distributing the content. Also, the content is divided into 10-second segments. Information (content acquisition destination, content type, resolution, maximum bit rate, DRM information) necessary for streaming distribution / playback of content is described in an MPD (The Media Presentation Description) file.

  The client analyzes the MPD file, sends a request for the corresponding content segment to the server as an HTTP GET REQUEST, and receives the content segment from the server (as an HTTP GET RESPONSE). When the available bandwidth of the client is narrow, the client selects a low-rate segment at the time of request transmission, and changes the encoding rate by requesting it.

  In Non-Patent Document 2, HTTP 2.0 (SPDY) is further applied to increase the initial congestion window size, multiplex requests (HTTP GET Request), and compress HTTP headers that are redundant. Has been realized.

IETF httpbis Internet-Draft, Hypertext Transfer Protocol version 2 (14) The 2013 IEEE International Conference on Multimedia and Expo (ICME) 2013, p1-p6, DYNAMIC ADAPTIVE STREAMING OVER HTTP / 2.0

  In Non-Patent Document 1, the initial congestion window size is extended to the maximum value of 64 Kbytes in the TCP connection specification. However, in order to realize multiplexing of HTTP sessions on a single TCP connection, the window size is compared with the method that artificially expands the initial congestion window size like existing HTTP / 1.1 that uses multiple TCP connections. However, it is not enough, and there is a problem that it becomes a limitation of speeding up Web content distribution.

  When changing the congestion window size of the server, it is necessary to change the server setting or to send a SPDY (registered trademark) WINDOW_UPDATE frame. The former case is not realistic because it is necessary to set one server at a time. In the latter case, there is a problem that a plurality of exchanges are required between the client and the server, and a large initial congestion window size cannot be used.

  Non-Patent Document 2 simply applies SPDY (registered trademark) to MPEG-DASH. Therefore, the same technical problem as in Non-Patent Document 1 is included.

  The object of the present invention is to solve all the problems of the prior art and establish multiple SPDY sessions between a client and a server to simulate the initial congestion window size of a single TCP connection without modifying the server. It is to realize high-speed Web content distribution that can be expanded.

  To achieve the above object, according to the present invention, in a Web content distribution method in which a server distributes content to a client through an SPDY session, the client establishes a preceding SPDY session with the server and requests a content configuration file. And the client establishes multiple subsequent SPDY sessions with the server using the waiting delay period from the request to establish the predecessor SPDY session until the configuration file is received, and the client establishes the predecessor SPDY session. Including a procedure for receiving each sub-content whose address information is described in the configuration file received in parallel in each SPDY session.

According to the present invention, the following effects are achieved.
(1) Since the initial congestion window size can be artificially increased using multiple SPDY sessions, it will be possible to expect faster web content delivery.

  (2) By executing a procedure to establish multiple SPDY sessions with the server during the response delay period from requesting the establishment of the preceding SPDY session to receiving the configuration file, multiple SPDY sessions are created. Since the time required for establishment is overlapped with the response delay period from the establishment request of the preceding SPDY session to the reception of the configuration file, the time required to additionally establish multiple SPDY sessions will delay content delivery. Can be prevented.

  (3) Since the delivery method of each segment in HTTP Streaming can be switched according to the transfer rate of the segment, it is possible to play without interruption even at low speeds, and it is possible to cope with sequence bar advance at high speeds. Improvement is possible.

  (4) Since it can be handled only by renovation and implementation on the client side without revising the server side, it can be easily applied to existing systems.

It is a figure for demonstrating the outline | summary of the content delivery method of this invention. It is the flowchart which showed operation | movement of one Embodiment of this invention. It is a sequence flow between the client / server when the content is a Web page. It is a sequence flow between the client / server when the content is HTTP Streaming. It is the figure which showed the 1st delivery method of the segment in HTTP Streaming. It is the figure which showed the 2nd delivery method of the segment in HTTP Streaming.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a diagram for explaining the outline of the Web content distribution method of the present invention. A client 1 establishes a plurality of SPDY sessions with a server 2 and configures Web content using each SPDY session. A plurality of sub-contents to be received in parallel.

  The Web content to be distributed by the present invention is composed of one configuration file and a plurality of sub contents, and the acquisition destination address information of each sub content is described in the configuration file.

  The configuration file is an HTML file if the Web content to be acquired is a Web page, and an MPD (Music Player Daemon) if it is HTTP Streaming. The sub-content is content data such as each image displayed on the page if the content to be acquired is a Web page, and each segment constituting a streaming video if HTTP Streaming.

  The client 1 first establishes one SPDY session (preceding SPDY session) with the server 2 in advance. Then, in the preceding SPDY session, a content configuration file is requested to the server [FIG.

  The client 1 further establishes a plurality of SPDY sessions (following SPDY sessions) with the server 2 during a response delay period from when the server 2 is requested to establish the preceding SPDY session until the configuration file is received. An additional establishment procedure is executed [Fig. (B)]. As a result, the time required to establish multiple subsequent SPDY sessions with the server 2 can be overlapped with the response delay period from the establishment request of the previous SPDY session to the reception of the configuration file, so multiple SPDY sessions are added. It is possible to prevent the time required for establishing the content from delaying the content distribution.

  Thereafter, when a plurality of succeeding SPDY sessions are established with the server 2, the client 1 that has received the configuration file establishes a preceding SPDY session that has finished sending and receiving the configuration file and a plurality of newly established succeeding SPDY sessions. Using the SPDY session, requests and reception of multiple sub contents described in the configuration file are executed in parallel [(c)]. As a result, the initial congestion window size can be increased in a pseudo manner, so that high-speed content distribution can be expected.

  FIG. 2 is a flowchart showing the operation of one embodiment of the present invention, FIG. 3 is a sequence flow when the Web content to be distributed is a Web page, and FIG. 4 is a case of HTTP Streaming. It is a sequence flow.

  First, referring to FIGS. 2 and 3, when a Web content distribution request to a client is detected in step S1, a SPDY session establishment request is transmitted to the server that stores the Web content (step S2). At time t0), by responding to the server, one SPDY session (preceding SPDY session) is established between the client and the server (time t1), and further from the client to the server, in the preceding SPDY session. Then, a GET request for receiving the content configuration file is transmitted (time t2).

  In parallel with step S2, in step S3, a procedure for establishing a plurality of SPDY sessions (following SPDY sessions) between the client and the server is started (time t3 to t4). In step S4, it is determined whether the web content requested for distribution is a web page or HTTP streaming.

  If it is a Web page distribution request, the process proceeds to step S5, and the process proceeds to step S6 after waiting for the requested configuration file (here, HTML file) to be received in the preceding SPDY session (time t5). In step S6, the address of each sub-content described in the received HTML file is identified, and a GET request message is generated.

  In step S7, each GET request is made using a plurality of subsequent SPDY sessions established with the server after the establishment procedure is started in step S3 and the preceding SPDY session in which the transmission / reception of the configuration file has been completed. Are transmitted to the server in parallel (time t6, t7, t8), and each sub-content distributed by the server in response to the GET request is received (time t9 to t14).

  In step S8, it is determined whether or not reception of all sub contents has been completed. If not completed, the process returns to step S7, and after transmission / reception is completed, an unsent GET request is transmitted using the subsequent SPDY session, and the sub-content responded by the server is received.

  Next, referring to FIGS. 2 and 4, if it is determined in step S4 that the request is an HTTP Streaming distribution request, the process proceeds to step S9, and the requested configuration file (here, MPD file) is received in the preceding SPDY session. (Time t15), the process proceeds to step S10.

  In step S10, the reception order of the segments described in the received MPD file is determined based on the playback order. In step S11, the address of each segment is identified, and a GET request message is generated.

  In step S12, a request for a segment file and a plurality of succeeding SPDY sessions established with the server after the establishment procedure is started in step S3 and a preceding SPDY session for which transmission / reception of the configuration file has been completed are used. The reception is performed in parallel.

  In the present embodiment, first, a GET request related to segment # 1 whose playback order is No. 1 is transmitted to the server (time t16), and segment # 1 distributed by the server in response to the request is received (time t17). .

  Next, the reception quality of each subsequent segment # 2, # 3, # 4 is determined based on the time required to receive the segment # 1 (time t18). Then, using a plurality of established SPDY sessions, each GET request including the reception quality designation is sent to the server in parallel (time t19, t20, t21), and the server delivers in response to the request The sub contents # 2, # 3, and # 4 are received with the specified quality (t22, t23, t24).

  In step S13, it is determined whether reception of all sub-contents has been completed. If not completed, the process returns to step S12, and the reception quality of the subsequent segments # 5, # 6 is determined based on the time required to receive the segments # 2, # 3, # 4 (time t25).

  Next, using a plurality of SPDY sessions for which transmission / reception has been completed, GET requests for unreceived segments # 5 and # 6 including the reception quality designation are transmitted (time t26, t27), and segment # 5 to which the server responds , # 6 are received (time t28, t29).

  5 and 6 are diagrams schematically showing a method for distributing each segment in HTTP Streaming. In the present embodiment, a reference speed Vref for switching the segment distribution method is set based on the relationship between the segment transfer speed Vi and the playback time ti, and depending on whether the segment transfer speed Vi exceeds the reference speed Vref or not. To change the transfer method.

  When the segment transfer speed Vi is lower than the reference speed Vref, as shown in FIG. 5, n continuous segments are divided into n (here 3 A distribution method is adopted in which one is received in parallel in each SPDY session and this is repeated as one cycle.

  On the other hand, when the segment transfer speed Vi is sufficiently fast and exceeds the reference speed Vref, as shown in FIG. 6, n × m continuous (here, A 3 × 3) segment is received in parallel for each of n SPDY sessions in parallel, and a distribution method is used in which this is repeated as one cycle.

  According to the present embodiment, since the initial congestion window size can be artificially increased using a plurality of SPDY sessions, it is possible to expect an increase in Web content delivery speed.

  In addition, according to the present embodiment, a procedure for establishing a plurality of SPDY sessions with the server is executed during a response delay period from when a client requests establishment of a preceding SPDY session until receiving a configuration file. The As a result, the time required to establish multiple SPDY sessions with the server can be overlapped with the response delay period from the request to establish the previous SPDY session to the reception of the configuration file, so multiple SPDY sessions are additionally established. It is possible to prevent the time required for the content from being delayed.

  Furthermore, according to the present embodiment, since the distribution method of each segment in HTTP Streaming is switched according to the segment transfer speed Vi, it is possible to perform continuous playback even at low speeds, and to advance the sequence bar at high speeds. It is possible to improve convenience such as being able to respond.

  Furthermore, according to the present embodiment, it is possible to cope with only the modification and mounting on the client side without modifying the server side, so that application to an existing system becomes easy.

  1 ... Client, 2 ... Server

Claims (4)

In the web content delivery method in which the server delivers web content to the client in an SPDY session,
A procedure for the client to establish a predecessor SPDY session with the server and request a content configuration file;
A procedure for establishing a plurality of subsequent SPDY sessions with a server using a waiting delay period from when a client requests establishment of the preceding SPDY session until receiving a configuration file;
A Web content distribution method comprising: a step of receiving each sub-content whose address information is described in a configuration file received by the client in the preceding SPDY session in parallel in each SPDY session.   2. The Web content distribution method according to claim 1, wherein the content is a Web page, the configuration file is an HTML file, and the sub-content is each content data displayed on the Web page.   2. The Web content distribution method according to claim 1, wherein the content is HTTP Streaming, the configuration file is an MPD file, and the sub-content is each time-series segment file. The procedure of receiving in parallel is as follows:
A first procedure in which n consecutive segments are received in parallel in each of n SPDY sessions, and this is repeated as one cycle;
Selectively adopt the second step of receiving consecutive n × m segments in m consecutive in each of n SPDY sessions and repeating this in one cycle,
4. The Web content distribution according to claim 3, wherein the first procedure is adopted when the segment transfer rate is slower than a predetermined reference rate depending on the segment reproduction time, and the second procedure is adopted at the earliest. Method.