Multiplex communications – Data flow congestion prevention or control
Reexamination Certificate
1997-11-24
2001-11-27
Hsu, Alpus H. (Department: 2662)
Multiplex communications
Data flow congestion prevention or control
C370S236000, C370S401000, C370S420000
Reexamination Certificate
active
06324163
ABSTRACT:
TECHNICAL FIELD
This invention relates in general to data processing networks, and in particular, to LAN emulation in switched ATM networks.
BACKGROUND INFORMATION
Asynchronous Transfer Mode (ATM) is an emerging network technology that is designed to transport information between communicating stations in a point-to-point fashion. The interest in ATM is its promise of high bandwidths and quality of service. ATM is a connection oriented architecture, in contrast to network architectures that are structured to broadcast data from the source to the destination. In ATM, the source negotiates a connected path to the destination before it proceeds to transmit its information to the recipient. ATM protocols (or rules, usually implemented in software) define the communications necessary to establish the connection. An ATM attached device has an ATM address in addition to any other network addresses it might have, depending on the particular ATM configuration within which it is incorporated. Some possible configurations will be described subsequently. Once a connection is established, the source station transmits its data only to the destination (a “unicast”).
In contrast to connection oriented architectures are broadcast networks. In these, data is sent from a source station to a destination station by broadcasting it to all addresses where the recipient plucks it off the network while the other stations on the network ignore traffic not bound for them. Broadcast architectures provide one motivation for structuring a network as a set of interconnected subnetworks or subnets.
In a large network, the proliferation of broadcast packets would overwhelm the network. Another simply reflects the nature of the pattern of growth of network communication generally. Although a particular network may start out as a freestanding Local Area Network (LAN), eventually end-station users will probably want to avail themselves of the services available on other networks, and look to connect their network with other networks. When this occurs, it is intuitive, as well as more precise, to view the resulting network structure as a set of subnets within a larger network, for example, an “internetwork.” However, a station on one internetworking subnet that wishes to communicate with a destination on another subnet can only do so if there is connectivity between the subnet in which the source resides and the subnet in which the destination resides.
Communications methodologies between subnets are usually termed to as “layer-3” protocols. This refers to the layered architecture networking model of the International Standards Organization (ISO). This model is illustrated in FIG.
1
. Layer-3 may sometimes be referred to as the “network” layer, and is equivalent to the “internetworking” layer in the TCP/IP model.
Connectivity between layer-3 subnets is provided by a device termed a router. When a source station on one layer-3 subnet wishes to communicate with a destination station on another layer-3 subnet, it broadcasts the data in the usual way. However, now it is the router that plucks the data packets off the first subnet and forwards it to the destination station via the destination station's layer-3 subnet to which the router is also attached.
Numerous types of networks coexist in the data communications industry. In addition to ATM, there may be LANs, Wide Area Networks (WANs), and others. There is a need in the industry for interconnection between different network architectures and, in particular, users of preexisting LANs have a need to connect to emerging high speed network technologies, such as ATM. The need for incorporating or interfacing preexisting networks (more precisely subnetworks) into an ATM environment has led to the specification of several methodologies to support preexisting network architectures within ATM.
One such methodology is the emulated LAN (ELAN) which simulates classical LAN protocols in an ATM environment. (Classical LAN protocols, for example Ethernet and Token Rings, are referred to as legacy LANs.) The protocols that provide the specification for ELANs are called LAN emulation (LANE). Layer-3 protocols run on top of ELANs just as they do in legacy LANs. Hosts attached to the ELAN include emulation software that allows them to simulate legacy LAN end stations. Such hosts are called LAN Emulation Clients (LEC). The LEC software hides the ATM from the LAN protocols within the LEC device, and a LEC can utilize those protocols as if it were a legacy LAN. A LEC can also provide a standard LAN service interface to a layer-3 entity in the same layer-3 subnet. Such a LEC is a LAN Switch that is usable to interface a legacy LAN with an ELAN.
Communication between LECs on an ELAN can be effected over the ATM. Each LEC has a physical, or Media Access Control (MAC) address associated with it, as well as an ATM address. For one LEC on a ELAN to communicate with another, it must obtain the ATM address of the destination LEC, given the destination MAC address. This address resolution is mediated through a LAN Emulation Server (LES). The source LEC issues a LANE Address Resolution Protocol Request (LE_ARP_Request) to the LES. Provided the destination station has previously registered its MAC address, ATM address pair with the LES serving the ELAN, the LES returns the ATM address of the destination to the requesting LEC in an ELAN Address Resolution Protocol Reply (LE_ARP_Reply). The source LEC can then use the ATM address to establish a connection to unicast data to the destination, a so-called data direct Virtual Channel Connection (VCC), and transmit its data to the destination thereon.
LANEs are also specified for emulation of source routed LANs, for example Token Rings, as well as nonsource routed LANs, such as Ethernets. In source routed LANs, packets being sent to a destination station contain a Routing Information Field (RIF) that includes a path from source to destination that is an ordered set of route descriptors, ring and bridge numbers, forming the route between source and destination station. Operations performed on MAC address described hereinabove are correspondingly performed on the RIF in a source routed ELAN.
Should the destination not have registered with the LES, the source communicates with the destination using conventional LAN methodology. This is mediated through a Broadcast and Unknown Server (BUS). The LEC sends its data to the BUS which then broadcasts it. Just as in a legacy LAN, the broadcast data is plucked from the network by the destination station, and is ignored by the other devices on the network. Exactly the same process is used if the destination is on a subnetwork, either a legacy LAN or an ELAN, in a different layer-3 subnet. In that case, the broadcast data is gathered by a router connected to the ELAN and forwarded via layer-3 protocols to the destination, as described hereinabove, just as if the ELAN were a legacy LAN.
When an LEC seeks to join an ELAN, connections must be established between the LES, BUS and the LEC. This initialization may occur when a particular LEC is powered up, for example, or if the LANE service or switch network recovers from a fault or is restarted. A LEC joining the ELAN must be added to point-to-multipoint connections that carry control and data frames to the LECs on the ELAN from the LES and BUS. Point-to-multipoint connections in an ATM network are connections from a single source station to multiple destination stations. Here, the LES and BUS each represent the single source station and the LECs are the multiple destination stations for the connections carrying control and data frames to the LECs. The process of adding LECs to these multipoint connections requires ATM switches in the network to process special control messages.
In large networks, switches may not have the processing power or memory to handle the addition of large numbers of LECs when they join an ELAN simultaneously. This can occur if the LANE service or switch network recovers from a fault or is restarted. If such a congested state occurs in th
Alexander, Jr. Cedell Adam
Ervin Jim P.
Rovner Edward Joel
Squire Matthew Blaze
Tippens Mark Alan
Cockburn Joscelyn G.
Ho Duc
Hsu Alpus H.
International Business Machines - Corporation
LandOfFree
LAN emulation service delay of multicast tree generation... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with LAN emulation service delay of multicast tree generation..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and LAN emulation service delay of multicast tree generation... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2600570