Half-covered self-organizing dynamic multicast routing method

Abstract

The invention relates to a routing method for realizing Internet multicast in a semi-covering and self-organizing manner, including a network topology organization method and a data packet forwarding control method. The network topology organization method dynamically reconstructs the network into hardware multicast domains, network layer multicast domains and overlay multicast domains according to the positional relationship and link characteristics among members of the same group. The connections are implemented in a single routing protocol. The datagram forwarding control method performs replication, encapsulation or conversion on the multicast data packet according to the established semi-coverage network topology, and then chooses among the three methods of link-level delivery, network layer routing and forwarding, and overlay forwarding. And the most efficient way is to complete the multipoint delivery of multicast data.

Description

Technical field: [0001] The invention relates to a computer network system structure, a method and a system for Internet multicast routing. Background technique: [0002] Multicast is a data forwarding and delivery mechanism that delivers packets to multiple destination sites at one time. A group ID represents a group of nodes in the network. Multicast simplifies the multipoint delivery process of the source, and more importantly, it can reduce the network burden by reducing link duplication packets and improve the efficiency of data delivery. [0003] According to the principle of protocol layering, multicast can be divided into hardware multicast, network layer multicast and overlay multicast. Hardware multicast is implemented at the link layer and physical layer, and the group identifier is the address of the link layer group, and its communication range is usually limited to the same physical subnet of the nodes. The essential feature of network layer multicast is tha...

AI Technical Summary

Problems solved by technology

Second, when the network topology and membership changes, the original forwarding tree will partially or completely fail and need to be recalculated. Therefore, the ability of IP multicast to adapt to changes is not strong. A small number of existing multicast applications are members and long-term fixed topology

Third, the overhead will multiply as the number of groups increases, so the scalability of IP multicast is poor

Fourth, because the scattered member nodes are difficult to generalize, IP multicast cannot effectively support advanced network features including reliable transmission, flow and congestion control, etc. like unicast

Fifth, in order to improve availability, the IP multicast protocol imposes more restrictions on the network environment. For example, PIM-DM is only suitable for relatively concentrated members. These assumptions limit the scope of use of the IP multicast protocol.

[0008] Although overlay multicast solves the main problems of IP multicast in a targeted manner, at the cost, it also introduces some defects that IP multicast did not have.

First, because overlay multicast does not understand the physical topology of the network, it cannot optimize the data distribution process according to the topology, so there are many duplicate packets on the link

Second, since there is no multicast logic at the network layer, it cannot directly invoke subnet hardware multicast support, which wastes resources and reduces efficiency

Third, overlay multicast needs to access transport layer and application layer protocol headers to decide how to forward packets. Therefore, compared with IP multicast, its packet processing process is more complicated, and the packet stays longer in intermediate nodes. long

Fourth, hardware multicast and IP multicast basically do not have any assumptions about applications, while the protocol layer of overlay multicast determines that it is closely related to applications. Most of the existing overlay multicast systems only target one or one type of application Optimized, poor versatility

These existing improvement schemes are still in a relatively preliminary stage, mainly in the following aspects: first, the coverage multicast and non-coverage multicast protocols they use are independent of each other, and the improvement schemes only increase the communication between different routing protocols. Therefore, the application of these schemes requires the deployment of multiple different multicast routing protocols on the same node. On the other hand, the interaction mechanisms in these schemes are not universal and interoperable.

Second, they assume that the boundaries of non-coverage multicast domains are fixed, that is, IP multicast domains and hardware subnets will not be dynamically merged or fragmented. This assumption is not consistent with the development status of the network environment. do not adapt

Second, there are various types of nodes, and their computing, storage, and communication capabilities vary greatly

At present, the research aimed at integrating the respective advantages of various multicast methods is still in a relatively preliminary stage, and the performance is as follows: 1) cannot have the ability of coverage forwarding and non-coverage forwarding at the same time in a single protocol; 2) cannot adapt to the new network environment Development changes and needs; 3) In the case of diversified access technologies, the scalability of multiple hardware multicast technologies has not been fully considered

Images