Multiplex communications – Pathfinding or routing – Switching a message which includes an address header
Reexamination Certificate
1999-05-18
2004-03-23
Vanderpuye, Kenneth (Department: 2661)
Multiplex communications
Pathfinding or routing
Switching a message which includes an address header
C370S401000, C370S432000
Reexamination Certificate
active
06711163
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to multicasting in data communication networking and, more particularly, to multicasting in a local area network (LAN) bridge.
Traditionally, a LAN was a single broadcast domain shared among network devices needing to communicate with one another. As more and more network devices were added to such LANs, competition for the shared bandwidth intensified and communication began to slow. To overcome this problem, devices called bridges were interposed to segment the network devices into multiple broadcast domains. Traditional LANs then began to be called “LAN segments” within a larger network topology which included multiple LAN segments, bridges, and often times a router for linking the network devices on the LAN segments with a backbone network.
Bridges are intelligent devices which transmit packets from one broadcast domain to another over a shared backplane. Rather than transmitting packets indiscriminately, however, network interfaces on a bridge apply Media Access Control (MAC) bridging rules. Each network interface learns the MAC addresses of the network devices connected to the bridge through itself from the source MAC addresses encoded packets received from such devices. When a subsequent packet is received on the backplane, the network interfaces are then able to individually check the packet's destination MAC address and determine if such address is among the list of previously learned MAC addresses. If the address is found in the list, the interface forwards the packet. If it is not found in the list, the interface filters the packet, unless no other interface on the bridge has claimed the packet. Through such “look up” operations conducted at each network interface, bridges advantageously reserve backplane and media bandwidth for packets requiring transmission in order to reach their intended destination.
Recent vintage bridges often implement bridging rules in custom integrated circuits. Such bridges are commonly referred to as “switches”, though their core function is not very different from traditional bridges. But by reducing basic bridging “look up” operations to custom logic, switches are able to pass traffic at or near wire speed. Switches thus have reduced packet loss and latency as compared with more processor-dependent bridges.
While bridged networks have advantages in speed and simplicity of implementation, they have experienced problems handling multicast traffic. Multicast packets are not destined for any one network device, so the destination MAC addresses encoded in such packets do not correspond to a “source learned” address on any one network interface. Absent additional rules, all network interfaces within the same subnetwork will therefore capture and indiscriminately retransmit multicast packets. Incident to such packet “flooding”, however, is consumption of bandwidth on media where no destination network device resides. In a multi-bridge environment, a spanning tree algorithm must also be run to prevent loops, imposing an additional tax on performance.
One alternative to flooding multicast packets is conventional multicast routing. Multicast routing restricts multicast traffic to particular subnetworks and therefore reduces flooding. Accordingly, while speed and simplicity of implementation generally points toward bridging, multicasting requirements suggest a continued place in a bridged network for routing. Out of this dichotomy were born bridges supporting both bridging and multicast routing. In a typical arrangement, a multicast router is configured on a processing entity logically interposed on a bridge backplane between network interfaces. This “internal” router learns the multicast groups to which each of the bridge's network interfaces belongs. Network interfaces transmit multicast data packets to the router using a well-known destination MAC address of the router interface. The router consults a multicast routing database and resolves the packet's multicast destination network address (i.e., multicast group address) to a set of network interfaces on the bridge supporting such address. The router then retransmits the packet on the backplane. The network interfaces then apply conventional MAC bridging rules to the transmitted packet. Through such “look up” operations performed at the router and network interfaces, a bridge/router advantageously reserves media bandwidth for only those multicast packets destined for subnetworks to which network interfaces belong.
The bandwidth savings realized by implementing multicast routing on a bridge has, however, come at a price. First, multicast routing has required an extra step in the forwarding process. Where a bridged packet is reviewed only at the network interfaces, a packet routed by a multicast router is reviewed at the network interfaces and a router interface. The added routing step is not only time-consuming but requires a second transmission across the backplane. Moreover, some LAN segment bandwidth is still wasted because multicast routers discriminate only at the interface level. Packets therefore continue to be flooded out ports associated with an interface which has a host belonging to the multicast group even where no network device belonging to that multicast group is reachable through the port. Accordingly, there is a need for a multicasting capability for a bridge which has superior speed and bandwidth conservation characteristics.
SUMMARY OF THE INVENTION
In one aspect, the present invention improves multicast communication in a bridge through the expedient of distributed multicast forwarding with sub-interface granularity. A plurality of network interfaces share a backplane of a bridge. The network interfaces each have a plurality of ports and retain a local multicast database which associates multicast groups active on the interface with local ports active in such groups. Forwarding decisions on multicast data packets transmitted on the backplane are made by network interfaces by referencing their local multicast databases. Each network interface forwards multicast data packets only on the local ports which belong to the multicast group identified in the packet.
In another aspect, configuration of the local multicast databases is assisted by a management interface which shares the backplane with the network interfaces. The management interface retains a global multicast database which associates multicast groups active on the bridge with ports active in such groups, The management interface updates the active port lists for multicast groups in the global multicast database with group/port association changes learned from multicast control packets transmitted on the backplane. The management interface transmits forwarding updates to the network interfaces having ports belonging to such groups.
In another aspect, a particular network interface is made responsible for each multicast group for forwarding multicast control packets and unknown multicast data packets to management interface for learning in order to reduce oversubscription of backplane bandwidth.
These and other objects of the invention can be better understood by reference to the following detailed description, taken in conjunction with the accompanying drawings which are briefly described below. Of course, the actual scope of the invention is defined by the appended claims.
REFERENCES:
patent: 4706080 (1987-11-01), Sincoskie
patent: 4740954 (1988-04-01), Cotton et al.
patent: 5331637 (1994-07-01), Francis et al.
patent: 5522042 (1996-05-01), Fee et al.
patent: 5608726 (1997-03-01), Virgile
patent: 5740175 (1998-04-01), Wakeman et al.
patent: 5790545 (1998-08-01), Holt et al.
patent: 5812771 (1998-09-01), Fee et al.
patent: 5898686 (1999-04-01), Virgile
patent: 6085238 (2000-07-01), Yuasa et al.
patent: 6091725 (2000-07-01), Cheriton et al.
patent: 6101187 (2000-08-01), Cukier et al.
patent: 6216167 (2001-04-01), Momirov
patent: 6331983 (2001-12-01), Haggerty et al.
patent: 6370142 (2002-04-01), Pitcher et al.
“IP Multicast Streamlines Delivery
Goodwin L. Michele Wright
Reid Steven William
Alcatel
Cordeiro David A.
Reader Scot
Vanderpuye Kenneth
LandOfFree
Data communication system with distributed multicasting does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Data communication system with distributed multicasting, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Data communication system with distributed multicasting will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3283984