Abstract
The IP-multicast transmission is the IP level answer for the growing one-to-many content spreading needs in multimedia applications (Hosszú, 2005). Nevertheless the address allocation and service discovery is a problematic field of this technology. Despite of the efficiency of the IP-multicast it has not been deployed in the whole Internet. Especially the global address allocation is a problematic part of the Internet-wide multicasting. This article addresses such problems in order to review the existing methods and the emerging research results. The IP-multicasting uses a shared IPv4 address range. In Internet-wide applications the dynamic allocation and reuse of the addresses is essential. Recent Internet-wide IP-multicasting protocols (MBGP/ MSDP/PIM-SM) have a scalability or complexity problem. The article introduces the existing solution for the wide-area multicasting and also proposes a novel method, which overcomes the limitations of the previous approaches.
TopBackground
The IP-multicasting provides an excellent solution for the one-to-many communication problems. The IP-multicast is based on the routers, which act as nodes of the multicast distribution tree. In such a way the routers multiply the multicast packets to be forwarded to every member of the multicast group. The IP-multicast method relies on network level mechanisms, since the construction of the multicast delivery tree is based on the multicast routing protocols in the routers. In such a way the IP-multicast is a pure network level communication technology, which is a logical extension of the unicast (one-to-one) IP-based communication.
The alternative of the IP-multicast is the application-level multicast (ALM), where the multiplication points of the multicast distribution tree are the hosts and not the routers as in case of the IP-multicast (Banerjee, Bhattacharjee, & Kommareddy, 2002). The ALM methods are inherently less efficient than the IP-multicast, since the hosts in case of the ALM generate duplicated traffic around the hosts. Another disadvantage of the ALM is the inherent unreliability of its multiplication points, since these are host, which are run by users without any responsibility for the whole communication.
The sophisticated IP-multicast routing protocols, such as the distance vector multicast routing protocol (DVMRP) (Thyagarajan & Deering, 1995), the most widely used protocol independent multicast—sparse mode (PIM-SM) (Fenner, Handley, Holbrook, & Kouvelas, 2006), and the experimental bi-directional protocol independent multicast (BIDIR-PIM) (Handley, Kouvelas, Speakman, & Vicisano, 2005) ensure that building and ending the multicast distribution trees has already been solved inside a routing domain, where all the routers are under the same administration (or a strict hierarchy of the administrators), where there is a homogenous infrastructure for registering the sources and the receivers. Here the uniform configuration of the routers is possible and the current routers automatically can enable the multicast traffic inside the domain, which means in practice the network of an autonomous system (AS). The sophisticated multicast routing protocols work efficiently inside a multicast routing domain; however, the Internet is composed of several Ass, and the wide-area multicasting needs the inter-AS (inter-domain) routing as well. Unluckily, the cooperation of the ASs in transmitting the multicast traffic has not completely solved yet. It has at least three reasons; one is the address allocation problem, but the source discovery and interdomain routing are still unsolved issues. All of them will be discussed in the following text.
The first problem is related to the topology among the ASs, where the IP-multicast traffic can be forwarded. Its reason is that oppositely to the IP-unicast traffic, the destination of the multicast packets is not single and normally cannot be determined based on the destination address. That is why the peering connections among the autonomous systems developed for the unicast transmission are not appropriate for forwarding the multicast traffic. A different interdomain routing is necessary.
Key Terms in this Chapter
IP-Multicast: Network-level multicast technology, which uses the special class-D IP-address range. It requires multicast routing protocols in the network routers.
Multicast Routing Protocol: In order to forward the multicast packets, the routers have to create multicast routing tables using multicast routing protocols. The most widely used multicast routing protocol is the protocol independent multicast (PIM).
Address Allocation: The problem of choosing an unused IP-multicast address before starting a multicast session, and when the session has been finished, this address should be released.
Autonomous System (AS): A network, where the main routers are in common administration. The Internet is composed of peering ASs, which are independent from each other.
Source Discovery: This problem arises when a host sends a join message to a router. The router can forward this join message toward the source of the multicast group if it has information about the source. This problem is more difficult, if the joining host and the source of the group are in different ASs. In this situation the MSDP can be used in order to exchange the information about the active sources among the ASs.
Inter-Domain Routing Protocol: IP-level routing protocol in order to create paths through the border-routers of the autonomous systems (ASs).
Domain Name System (DNS): Hierarchical distributed database for mapping the IP addresses to segmented name structure and vice versa.
Multicast Source Discovery Protocol (MSDP): This protocol makes it possible to use independent multicast routing inside the domains, while the multicast sessions originated from or to another, domains reach all the participants. In each AS there is at least an MSDP protocol entity in order to exchange the information about the active sources among them.