KR20040076540A - Implementation Scheduler for Multicast Transmission - Google Patents

Abstract

PURPOSE: A method of realizing a scheduler for multicast transmitting is provided to immediately set a group if users want to receive a service at a desired time, and to perform a scheduling process for the same group if other users request the same item, thereby simultaneously supplying the service to plural users with one channel. CONSTITUTION: A scheduler stores a number of a desired item, a user ID, and a multicast IP necessary for a service. If a user requests the service, a service socket of the scheduler generates one thread per request, and maintains each connection setup. A multicast group IP generated by the scheduler is transmitted to the user and a media server. The media server supplies the service to a multicast group. Clients automatically join in the multicast group by using the multicast group IP transmitted from the scheduler, and watch a requested video.

Description

Implementation of a Scheduler for Multicast Transmissions {Implementation Scheduler for Multicast Transmission}

The system implemented in the present invention was produced with Java JSDK version 1.4 and Java FrameWork (JMF). When the components are divided for the convenience of system management and program, the user checks whether there is a client requesting service to the server, whether there is an item already scheduled by receiving the client's request, and whether or not there is a new scheduling according to the result. The programming structure is structured into three parts, including the scheduler that determines the network controller, and the media server responsible for data transfer using the multicast group address and port number generated by the scheduler.

The user and the scheduler make sockets with each other, prepare to communicate, process the request sent by the user, and store the desired item number, the user's IP, and the multicast IP needed for the service in the scheduler. Since the user and scheduler are not one-to-one but n-to-one, when the user makes a service request, the scheduler's ServiceSocket generates one thread per request to maintain each connection. . The multicast group IP generated by the scheduler is delivered to the media server as well as the users, and the media server provides services to the multicast group, where clients automatically multicast group using the multicast group IP sent from the scheduler. Join to watch the requested video. At this time, the service is provided without delay when requested by the user.

The present invention provides a method for minimizing the load of a media server and using network resources more efficiently to provide a high quality multimedia service immediately by a user's service request to a server of a lower price.

As technology has advanced, researches and attempts to provide better services to users have continued, and as well as software necessary for data transmission, hardware advancement has made it possible to provide high-speed Internet transmission service and large data transmission service. High quality service was obtained. On the other hand, with the development of technology, users want to be provided with a high-speed service of large-capacity multimedia data including moving pictures, and a service that can satisfy various needs is required. In particular, requests for user-requested services, which were not supported by general broadcasting media, have increased significantly over time, and technology development for them has been required.

User-requested services are not only aimed at providing users with data services but are being transformed into user-oriented on-demand services by allowing them to receive the information they want at the time they want, at the desired quality. There is a demand for multimedia data that integrates various media such as video, music, and images. As a result, network networks need to transfer large amounts of data at high speed.

To realize an on-demand system that can service in real time when a large amount of data such as multimedia is required, various technical and cost problems must be solved. In particular, in order to provide a smooth service to users, a powerful processing capacity of a server that stores and manages data is required, and various studies have been conducted to solve problems such as the construction of a high-speed broadband transmission network capable of transmitting large amounts of data smoothly. It's happening. Therefore, a method using a multicast service using a distributed storage system handled in a CDN (Contents Delivery Network) or the like and grouping users who apply for the same data into a group and having a single resource at the same time has been studied.

With multicast, unlike traditional unicast transmission schemes, a service is provided to only one user with one resource, as well as the number of users who have requested data at the same time as the server load. Resource savings However, in the conventional multicast transmission method, the same group ID is determined and the service start time is announced through e-mail and the like, and the service is accessed. However, the system proposed by the present invention provides a method for an item desired by a user. When a service request is made, a service is delayed for a predetermined time to provide a service by grouping user groups requiring the same item in real time, thereby providing multicast transmission while providing user-oriented service.

Currently, a service mainly provided by the Internet uses a broadcast transmission method that waits for data at a predetermined time or a unicast transmission method that performs one-to-one transmission when a user request occurs. The reason why it is difficult to service using the multicast transmission method is that the service request time of the user requesting the item is not the same. Therefore, there is a need for a method that can provide services to black users at once using a single channel when many users request the same item within tens of seconds, such as a recently released movie.

1. Service Scheduler

Comparing the data transmission using the unicast transmission method and the multicast transmission method, the efficiency of the channel required to service the same item becomes more pronounced with more users. When the unicast transmission method is used, as many channels as the number of users who request the service are needed. However, when the multicast transmission method is used, one channel can be used. Therefore, the more desire for the same item, the more efficient use of resources. However, the service using the existing multicast transmission has decided the transmission time in advance in the server rather than receiving the service at the desired time, and allows the clients to access the service at a fixed time through e-mail or notification. Therefore, since the data is transmitted from the service provider's point of view, the on-demand service cannot be provided.

However, the proposed system is a system that provides on-demand service characteristics by eliminating the disadvantages of existing multicast, and the scheduler uses a single channel to service users who simultaneously request the same data in a service using a multicast transmission method. In order to be able to provide a multicast address for each group generated as a result of scheduling a user request, as shown in FIG.

The condition for the service request of users belonging to the same group by the scheduler is that the user must apply for the same item, and the multicast group setting time set to the wait time that the users can understand (about 20 seconds to 1 minute). Users must apply for services within the scope of). The group setting time is determined by analyzing the popularity of the items being serviced by the server and the user's request pattern. If the group setting time is set longer, the service can be provided to more users by using a single channel in one group, but users who apply for the service at the beginning of the group setting time have to wait longer.

The scheduler first accepts the user's request and decides which group the user should be served with. First, check the system resources to see if the user's request can be served. Since the capacity of the channel available in the system is fixed, the user's request should be rejected or waited when there is no space available for the service. If the system resources are sufficient to provide the service, the user who waits for the service already received by another user who selects the same item as the requested data is requested before the group setup time has elapsed from the service request time of the user who first requested the service. A user who requests a service for the same item before a group setting time elapses from a service request time of one user, and when another user requests the same item, the service is provided with the same multicast group address and port.

2 shows a process of grouping users' requests for a single item for a group setting time and serving a group unit using a multicast address.

As shown in FIG. 3, when performing services using the unicast transmission method, if n people request the same item from the user who requested the service to the user who requested the service before the end of the group setting time, Each of n channels should be allocated, but a single channel can be serviced by assigning a multicast group address.

Grouping requires user information on an item and information related to the item. User information includes the Internet address and the unique ID of the requested item, and the item information required for grouping stores the multicast address storage array to be used by the currently scheduled thread and the information to determine whether the multicast address generated by the function is available. Array that stores information to determine whether the generated port is available, a timer that executes a task that gathers users' requests during group setting time, and stops grouping after this time to start setting up a new group. have.

First, in order to distinguish whether an item requested by a user is currently scheduled or an item that needs to be newly scheduled, it has an array for storing information for determining whether grouping with an item ID as an index is in progress. When this scheduler receives a user request, it looks first and if you are scheduling it, refer to the array described below to refer to the multicast address and port currently being scheduled. An array with an item unique ID as an index and having a multicast address to be used for the item being scheduled as a value is referred to the user who has already requested the item being scheduled by referring to the array. The client then establishes a connection with the media server at this address. On the contrary, if there are no threads grouping to service the requested item, the newly created multicast address must be checked to see if it is already in use. Therefore, an array of whether or not to use the multicast address that is likely to occur is required. Similarly, it also needs to manage the ports to use, so it has an array of port usage flags. In addition, there is a timer that manages the entire grouping time and manages the service session while collecting user requests during the group setting time.

Therefore, the scheduler must maintain and manage the list of currently provided items, the information for management and the information of all users who requested the service, the timer for each item, and the list of users who are being served in the same group by movie. .

The system proposed in the present invention provides a service having an on-demand function by using a multicast transmission method, thereby increasing efficiency by transmitting a service to more users with one multicast group address.

2. Media server on demand

The media server stores a large amount of data requested by the user, transmits multimedia stream data to the user in real time when requested by the scheduler, maintains a connection with the client, and disconnects when the service is finished. The server needs a large data storage device to find and transmit appropriate data among various items according to the user's needs, and strong processing power is required to smoothly provide a large number of user requests.

In the method of establishing a connection with a user in the server, when using a conventional unicast transmission, since the service was provided to a single user with one resource, the Internet IP address was assigned to the destination field. When the transmission is used, since the service is provided to a plurality of users with one resource, the multicast IP allocated for the multicast transmission is assigned to the destination address field. Multicast IP for a group of users reduces the load on the server by generating one multicast address per group to group and service the requests of users who have requested the same data during the group setup time when the user requests from the scheduler. Brings effect.

The multicast group address used by the server to serve users uses the group address generated by the scheduler. If the scheduling is performed on the server to transmit the item requested by the user, all the servers perform different scheduling when the system needs to be expanded by using more than one server in the case of serving multiple users later. As a result, system integration will be difficult.

The existing service makes a direct request when the client requests a service request for an item desired by the client. However, in the case of the system proposed in this paper, the server does not need to know the internet address of the clients to service because the request is made through the scheduler. As you can see in Figure 1, the client's request is entered into the scheduler to see if a thread is being scheduled for the requested item, creating a new group address, or receiving a multicast group address already grouped. Therefore, as in the conventional method, there is no need for the client to enter the Internet address of the server that the client wants to request in the datagram packet or to manage the Internet address of the client to be provided by the server.

When the client is notified of the multicast group address to which it should subscribe, it closes the datagram socket previously associated with the scheduler and creates a new multicast socket to connect to the server. Since the client connects through a multicast socket, the server assumes that the user will join and is dedicated to the transfer.

3. Client

The client receives the data stream sent over the network and decodes it to the user.

In order for the client to request a service, it must request the desired item from the scheduler. If the data requested by the user exists on the server, the scheduler generates a multicast group address and informs the client and server. When the client gets a multicast group address from the scheduler, it closes the socket that was connected to the scheduler and creates a new multicast socket for communication with the server. In addition, the server creates a datagram socket to send data to the client using the multicast group address, and the client enters the multicast group address received from the multicast socket in the source address field and waits for reception. , The server starts the transfer.

When communication is initiated, the client and server are serviced with a single common multicast group address without having to maintain direct information about each other. This is because you subscribed through a cast socket.

The client component of the system implemented using the new scheduling does not maintain a session with the server main program while receiving data from the server after receiving the multicast group address from the scheduler, and is responsible for the server threads that are assigned to each group. Maintain a session with. Because you can separate each connection into threads and improve channel management efficiency, when the service you want to receive ends, you automatically destroy the socket of the user and the server thread that sent the item is also terminated.

4. Terminal network

Currently, the environment in which multicast transmissions can be used smoothly is limited to the local network. Recently, various protocols have been developed to provide services using multicast group addresses, and routers that need to perform these protocols have been developed to support multicasting, but most routers on the Internet still do not support multicasting. It's a pre-supported piece of equipment, so it's going to take a lot of time to replace it. Therefore, the packet using the multicast group address limits the TTL value of the data packet from 1 to 3 so as not to leave the LAN network.

The most optimal environment for a system using the benefits of multicast is a user-dense structure that is geographically contiguous and multiple clients are connected to the same router. For example, it may be a video or movie service in an apartment complex, a case where education contents are designated and provided in a school LAN network, or a company provides educational materials in-house. However, if the infrastructure to expand the range of multicast transmissions is established on the Internet, it will be expanded to a large service network capable of streaming nationally the last program which has been popular in the existing broadcasting stations, or the release movies which attracted huge audiences. Will be.

1 shows Components of scheduling system

2 is a grouping interval of grouping session

3 is a required channel for unicast and multicast

The group scheduling system using multicasting proposed in the present invention provides a method for providing a higher quality service to a plurality of users with data requested by a user without significantly increasing the load on the server. In particular, a pre-processing process is needed to group users who request data while gathering user requests into a service group and provide services. Each element includes a media server, a scheduler, a client, and a network that are needed to facilitate management. It is divided according to the function in charge of the.

In order to implement the proposed multicast on-demand multimedia server, the MOD system implemented through the new multicast scheduler provides users with service requests through the same channel at small intervals in the environment where the user's needs are concentrated on specific popular items. In order to provide a service, a group of subscribers can be defined for a predetermined waiting time, and scheduling can reduce the load of a server and the use of a network channel by sharing a multicast group address and a port with a group of users.

This enables more efficient use of scarce network resources, and at the same time significantly reduces the load on the server, reducing the number of channels associated with service requests to 54.26% when using a unicast transmission scheme when clients order the same items. Can be. As a result, the performance of the system can be more than doubled and the cost of implementing the system can be significantly reduced because fewer traffic is sent over the server and the network.

Claims (2)

Scheduling technology that generates multicast group addresses and port numbers immediately upon user's service request and sends them to users and media servers Technology that allows users to join a multicast group automatically and provide services without any intervention