KR20040076660A - Parallel information delivery method based on peer-to-peer enabled distributed computing technology and the system thereof - Google Patents

Abstract

The present invention relates to a method and system for transferring information over a network using a transport protocol such as HTTP, FTP, or the like. The information transmission method and system according to the present invention divides the information to be processed and transmitted by the computer (server) that transmits the information into small pieces, and stores the duplicated / distributed data in a plurality of personal computers (PCs) connected to an arbitrary network. In this case, a peer-to-peer that transmits information to a computer (client) on behalf of these personal computers (PC) should be responsible for the information transfer that the computer (server) that transmits the information by using distributed computing techniques. Peer-to-Peer) characterized in that the transmission. According to the method and system for transmitting information on a network according to the present invention, it is possible to reduce the workload of a computer (server) that transmits information, and the computer (client) that receives the information can collect information in parallel through various paths. This can reduce the time it takes to receive information.

Description

PARALLEL INFORMATION DELIVERY METHOD BASED ON PEER-TO-PEER ENABLED DISTRIBUTED COMPUTING TECHNOLOGY AND THE SYSTEM THEREOF}

The present invention relates to a method of delivering information using distributed computing and peer-to-peer communication techniques over a network. More specifically, the present invention provides a unit of information (for example, image information such as JPEG or moving image information such as MPEG) divided into units such as a file or information consisting of data of various units (for example, a plurality of A web page composed of GIF files, etc.) is divided into data smaller than the original size and transmitted in parallel from a plurality of computers to a client.

In general, a computer (hereinafter referred to as a "client") requesting certain information in order to receive arbitrary electronicized information through any network and a computer providing a requested information from a client (hereinafter referred to as a "server"). ) And a network constituting an information transmission path between them. Since the server and the network are not resources that are serviced by only one client, they are heavily loaded by the demands of different clients, and the increase in the load eventually causes a bottleneck. As a result, there is a problem that the number of clients that can connect to the server at the same time is limited, and only a part of the bandwidth allocated to the client requesting information can be used.

Due to this problem, content providers who sell information through the Internet purchase large numbers of expensive and expensive computers that can act as servers to increase the number of users and improve the quality of service. Major investments are needed, such as increasing the bandwidth of the delivered network. In addition, content providers should anticipate and prepare for the maximum number of clients in order to stably provide a predetermined information providing service. Such a number of clients are difficult to accurately predict, and the number varies over time. If the information request does not occur as much as the estimated maximum number of clients, it means that the resources prepared by predicting the maximum number of clients are wasted. If the forecast is incorrect, the number of clients requesting information is estimated. In more cases, there is a problem that can not provide a normal service.

In order to solve the above bottleneck problem in the server and the network, a technique called IP multicasting has been devised. This is a technique to reduce the load on the server and the network when multiple clients require the same information at the same time, such as live video streaming through the Internet. In general, a unicast Internet application over TCP / IP sends a packet by the sender of the data knowing the recipient's IP address and marking the recipient's address in the header of the transport packet. In order for these packets to be delivered to the correct recipients, many routers on the Internet look at the headers of the packets to determine the transmission path. However, the configuration of the packet for multicast transmission is slightly different. The sender of the packet transmits the packet by indicating the group address in which the receiver participates, instead of the address of the receiver. The group address for multicast transmission is D-class IP address (224.0.0.0 to 239.255.255.255), which is a real host unlike A, B, and C-class IP addresses that represent individual Internet hosts around the world. It is not an address. Therefore, the receiver receiving the multicast packet determines whether the packet belongs to the group of packets or not. In order to use this IP multicasting technique on a real network, a function that supports IP multicasting is essential for router equipment. Only a few router equipments installed in the world currently support IP multicasting and Since there is no possibility that the router equipment located on the network will be completely replaced, there is a problem that the performance gain through the introduction of the IP multicasting technique is hardly expected under the current backbone network.

Another way to solve bottlenecks in servers and networks is to use a Content Delivery Network (CDN). CDN distributes contents through cache servers strategically installed in each network to provide users with fast and stable network access environment and to distribute users to Internet service providers such as ISP (Internet Service Provider) and content providers through distributed processing. It can be said to provide a network infrastructure. Content is delivered from the content provider's server to the user via the network or subscriber network. In this process, the slowdown of the Internet occurs in the "Middle-Mile", that is, the section from the content provider's server to the NSP or ISP's network to the user node. The CDN is the problem in this "mid-mile". In order to solve the problem, strategically installing a cache server in an ISP (Internet Service Provider) and distributing content from the CP server to the cache server has the same effect as the transmission in the same subnetwork rather than the transmission between networks. This ensures high speeds by preventing traffic from concentrating on specific networks. This CDN method is implemented by placing a computer in some important place on the network in place of the server to handle the server's request, thereby reducing the load on the server and the network until the client's request reaches the server. This can reduce the burden on the network by allowing them to be processed in the middle instead of crossing. However, the use of such a CDN also requires the cost of having a large number of computers comparable to those of the server, and even if the functions of the server are replaced by several computers, the number of servers serving the CDN function is limited. At this time, there is no fundamental solution to the problem described above.

SUMMARY OF THE INVENTION In view of the above-described conventional problems, the present invention provides a distributed processing technique that utilizes idle resources such as CPU, hard disk, and network bandwidth of a normal user's computer (hereinafter, referred to as "peer"), such as a PC. By using the peer-to-peer technique that allows data to be transferred between end users' PCs without going through the server, the information can be transmitted in parallel, improving the speed of information transmission on the network such as the Internet, and the load on the server and the network. The purpose is to reduce the number of channels and to shorten the path over which data is transmitted.

More specifically, the present invention divides and duplicates information to be serviced by a plurality of peers in place of a central server that is solely responsible for storing and transmitting information that is a target of a service, and stores information to be stored from a client to the information. When a request is made, the peers search for peers with fragmented data corresponding to this information, select peers that can achieve the fastest transmission speed, and send the information in parallel to the requesting client. In addition to being available and reducing the load on the server and network, the aim is to shorten the transmission path by transmitting data from adjacent peers rather than a central server.

1 is a conceptual diagram illustrating a schematic configuration of a system for transmitting information in parallel on a network according to the present invention.

2 is a block diagram illustrating the relationship between actual elements to which a system for transmitting information in parallel on a network according to the present invention is applied;

3 is a flow diagram illustrating the steps of a method for parallelizing the transmission of information on a network in accordance with one embodiment of the present invention.

In order to achieve the above object, an information transmission parallelization method on a network according to the present invention comprises the steps of: dividing and storing the information into data of a predetermined size; Searching for peers capable of providing the data of the predetermined size; Selecting a peer from among the searched peers according to a predetermined criterion; Receiving data of the predetermined size in parallel from the selected peer; And restoring the received data having the predetermined size to the information.

In addition, in order to achieve the above object, a system for transmitting information in parallel on a network according to the present invention includes a client requesting the transmission of the information; A server dividing and storing the information into data having a predetermined size; A plurality of peers storing the partitioned predetermined size data; And a peer suitable for transmitting the information among the plurality of peers according to a predetermined criterion by managing the data of the predetermined size and / or information about the data stored in the peers, and analyzing the information request of the client. It characterized in that it comprises one or more agents for selecting.

In addition, according to another aspect of the present invention, selecting a peer suitable for the information transmission may be performed in the client.

The features and advantages of the present invention for achieving the above objects will become more apparent from the following description taken in conjunction with the accompanying drawings. Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram showing a schematic configuration of parallel transmission to which the present invention is applied, and is composed of the following modules capable of transmitting information to each other through an arbitrary network.

A server (S) for dividing information to be transmitted and storing it as data of a predetermined size;

A client C for requesting predetermined information;

Peers (1, 2, 3, 4) (11, 12, 13, 14) which serve to transmit data of the predetermined size on behalf of the server (S);

One or more agents (A, B) 21, 22 for analyzing the client's request for information and retrieving peers that transmit the data of the predetermined size on behalf of the server in response to the client's request.

The server S divides and stores predetermined information into data of variable or fixed size. The divided and stored data may be encrypted. Such data encryption may be performed according to well-known encryption methods such as RSA (Rivest, Shamir, Adleman) method or DES (Data Encryption Standard), which are well-known public key encryption algorithms. This encryption technique may be particularly useful for transmitting paid content to clients.

At least one agent (A, B) (21, 22) distributes the data divided into the predetermined size stored in the server (S) to be stored in duplicate to each peer, and the divided It manages data, and discharges and imports the partitioned data, thereby constantly monitoring what peer data is stored in which information. In addition, the agent manages the monitoring result, and when there is an information request of the client C, the agent analyzes the information request and searches for which peer it can process.

At this time, when there is a request for information from the client C, the request may not be transmitted to the server S as in the conventional data transmission method, but to one or more agents A and B 21 and 22. have. In this case, one or more agents (A, B) 21, 22 store the requested information in the form of data divided into which peers 1, 2, 3, 4 (11, 12, 13, 14). It is determined whether or not it is desirable to transmit the divided data from among the peers according to predetermined criteria. The selection of one or more peers from which a plurality of peers is desirable for transmission may be performed according to the following criteria.

(1) By analyzing the client's IP address, a peer having an IP address similar to the client's IP address can be selected. For example, if the first three bytes of the IP address of the client and the peer are similar, the information transmission efficiency may be good since both are likely to be located on the same or adjacent network.

(2) The response speed of the peer can be considered to be good by using the Ping or Trace command. If the response speed is fast, it may be considered that the network load to the peer is not high, so the information transmission efficiency may be good.

(3) The CPU and I / O load of the peer can be measured using a dedicated application installed on the peer side, and a peer with less CPU and I / O load can be selected based on this. If a peer's CPU and I / O load is low, the peer can process the transfer more quickly, which can improve information transfer efficiency.

(4) The peer can be selected by examining the service history of the peer that has served the client.

Based on some or all of the criteria exemplified above, the agent (A, B) 21, 22 or the client C selects a peer having the desired transmission efficiency. The server S may also be selected according to the predetermined criterion as one of the peers. The selected peers transmit the data divided into their predetermined size to the client C. According to an embodiment of the present invention, if a failure occurs on the transmission network during transmission of divided data at each peer, a predetermined dedicated application running on the client may report the failure to a predetermined time-out. Can be detected and notified to the agents (A, B) 21, 22. In this case, the agent may select another peer suitable for transmission with the partitioned data and operate to send the partitioned data to the client C that the selected peer could not transmit due to a network failure. When the transmission of all the divided data is completed, the client C may restore the received divided data to its original form (which may include decoding depending on encryption) and use it for a purpose.

The system for transmitting information in parallel on a network according to an exemplary embodiment of the present invention may further include a cache manager that manages caching of the small unit of data transmitted between the peers. Such a cache manager caches data mainly transmitted between peers, and when the neighbor client requests this information, the cache manager directly reduces the load on the network. The cache manager may exist in the form of an independent server, but depending on the implementation, the cache manager may exist in each peer and report to the agent whenever a change in the cache occurs.

In this way, one piece of information is divided into data of a predetermined size and stored in different peers 1, 2, 3, and 4 (11, 12, 13, and 14), and the client C requests the information. In this case, each peer (1, 2, 3, 4) (11, 12, 13, 14) transmits the divided data of the predetermined size to the client C, respectively, so that the client is parallel from multiple computers (peers). You will receive a transmission. In a preferred embodiment according to the present invention, when the requested information is divided into N pieces, the client may receive transmissions in parallel from a maximum of N peers, but at the same time considering the characteristics of the information or the characteristics of the client, etc. It is also possible to include a function to limit the number of pieces of information to be a certain number or less.

FIG. 2 is a diagram illustrating the interrelationship between the server S, the client C, the agent 21, and the peer 11, which are actual components in the field to which the present invention is applied. It works like any of the flow charts shown in FIG. 3 below via any network.

3 illustrates a flow of information transmission parallelism on a network according to a preferred embodiment of the present invention.

In operation 310, the server S divides and stores all information that is a service target into data of variable or fixed size. At least one agent 21 shown in FIG. 2 distributes the divided data stored in the server S to each peer 11 and distributes the divided data stored in each peer 11. It manages and deletes the partitioned data and imports the new data. By monitoring the peer, the partitioned data of which information is stored in which peer. In addition, the agent identifies the information uploaded or deleted on the server S, and deletes the data stored in each peer 11 corresponding to the information deleted in the server S, corresponding to the newly uploaded information The divided data are distributed to each peer 11 redundantly.

In step 320, the agent manages the monitoring result, and uses this to analyze the information request when there is an information request from the client C and search for which peer to process.

In step 330, when there is a request for information from the client C, the request may be transmitted to the one or more agents 21, rather than to the server S as in the conventional data transmission method. In this case, the at least one agent 21 searches for which peers the requested information is stored in the form of divided data, and notifies the retrieved peers to the client C, and the client meets the above-mentioned predetermined criteria. Accordingly, from among the peers, it is determined from which peers it is desirable to transmit the divided data to select appropriate peers. As described above, according to the present invention, since the divided data are repeatedly stored in the peers, one or more peers having the same data exist on the network. Step 330 is a step of selecting a peer which is determined to have the best transmission efficiency among the one or more peers according to the above-described predetermined criteria. According to another example of the invention, selecting suitable peers may be performed at client C, but the agent may operate to select a suitable peer and notify client C of the selected peer. As described above, the server S may also be selected as one of the peers according to the predetermined criterion.

In step 340, the peers selected in step 330 transmit the divided data stored in the client to the client. Conceptually, if the information requested by the client is divided into N fragments, it can receive transmissions from up to N peers in parallel.However, considering the characteristics of the information or the characteristics of the client, the number of information fragments simultaneously delivered is fixed. It is possible to include a function that limits to less than a few.

In step 350, if the network between any peer and the client fails in the transmission process of step 340, or if an unexpected situation (such as a downtime of the peer computer) occurs, the transmission of the divided data that was performed in any peer is performed. Slow down or stop. In this case, the dedicated application running on the client side determines whether to time-out after a predetermined time elapses and notifies the agent that the transmission from any peer has slowed or stopped below a predetermined speed. If there is such a notification, the process returns to step 330 and the above-described process is repeated. In the case of such an unexpected accident, the client may notify the agent. However, the agent checking the transmission status may directly detect this and return to step 330.

In step 360, the transmission is complete when all the divided data from all peers is transmitted.

In step 370, the partitioned data transmitted is restored on the client side again. In this case, the data encrypted through a specific encryption algorithm is decoded again and restored to information of a form that can provide a predetermined service.

As described in detail above, according to the present invention, by dividing the information to be serviced by the server into a plurality of predetermined size data stored in a number of general user computer content providers to the external computer rather than the server that they manage It is possible to exclude the possibility that information may be leaked and used without the permission of the content provider.

In addition, according to the present invention, not only the one-to-one transmission between the server and the client, but also parallel transmission between a plurality of peers and clients, it is possible to dramatically shorten the time required to transmit information of the same size, Unlike the central server managing and serving requests centrally, since multiple peers process them in parallel simultaneously, the load on the server can be dramatically reduced, thereby reducing the cost of building the server.

In addition, the preferred embodiment of the present invention described above is for the purpose of illustration, it will be apparent to those skilled in the art that various modifications, changes, additions, etc. are possible within the spirit and scope of the present invention described above. Therefore, the scope of the present invention is defined by the claims described below, it will be seen that modifications, changes, etc. belong to the equivalent scope of the present invention. In particular, the above-described components of the server, the agent, the peer, the client, and the like are only functionally classified, and their actual physical location is irrelevant to the functional classification.

Claims (2)

In the information transmission parallelism method on a network, (a) dividing and storing the information into data having a predetermined size; (b) searching for peers capable of providing the data of the predetermined size; (c) selecting one or more peers among the searched peers according to a predetermined criterion; (d) receiving data of the predetermined size in parallel from the selected one or more peers; And (e) restoring the received data of the predetermined size to the information; Wherein the step (a) comprises Encrypting the divided data; And Controlling the data to be distributed and redundantly stored in the server and the peers. Information transmission parallel on the network, characterized in that it comprises a. In a system for transmitting information in parallel on a network, A client requesting the transmission of the information; A server dividing and storing the information into data having a predetermined size; A plurality of peers storing the partitioned predetermined size data; And A peer suitable for transmitting the information among the plurality of peers according to a predetermined criterion by managing the data of the predetermined size and / or information about the data stored in the peers, analyzing the information request of the client One or more agents to select Including, The predetermined criterion includes at least one or more of the workload of the peers themselves, the state of the network, the geographic location of the peers, the past service history, and the like. .