Multiplex communications – Pathfinding or routing – Switching a message which includes an address header
Reexamination Certificate
1998-10-19
2004-03-23
Hsu, Alpus H. (Department: 2665)
Multiplex communications
Pathfinding or routing
Switching a message which includes an address header
C370S408000
Reexamination Certificate
active
06711171
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to various methods and apparatus which provide distributed connection-oriented services for switched data communications networks, the service provided being scalable, allowing fully active mesh topologies, reducing broadcast traffic, and enabling connections to networks and servers outside the switch domain.
BACKGROUND OF THE INVENTION
Most data communications networks today rely heavily on shared-media, packet-based LAN technologies for both access and backbone connections. These networks use bridges and routers to connect multiple LANs into global internets.
A router-based, shared-media network cannot provide the high bandwidth and quality of service required by the latest networking applications and new faster workstations. For example, multimedia and full-motion video applications consume large amounts of bandwidth and require real-time delivery. Another high bandwidth application involves transmission of X-ray and other diagnostic information to enable doctors in different locations to consult while accessing the same patient information. Yet another application is “collaborative” engineering, i.e., allowing multiple engineers to work on the same project simultaneously while at different geographic locations. Thus, networks once used primarily for sending text files and E-mail or sharing common databases, are now being pushed to their limits as more users push more data across them.
One way to provide additional bandwidth on a given network segment is with larger shared-media pipes, such as FDDI or Fast Ethernet; however, this does not enable the application of policy or restricted access to the enhanced network resources. Alternatively, a network can be further segmented with additional router or bridge ports; however, this increases the cost of the network and the complexity of its management and configuration.
Switched networking is a proposed solution intended to provide additional bandwidth and quality of service. In such networks, the physical routers and hubs are replaced by switches and a management system is optionally provided for monitoring the configuration of the switches. The overall goal is to provide a scalable high-performance network where all links between switches can be used concurrently for connections.
One proposal is to establish a VLAN switch domain—a VLAN is a “logical” or “virtual” LAN in which users appear to be on the same physical (or extended) LAN segment, even though they may be geographically separated. However, many VLAN implementations restrict VLAN assignments to ports, rather than end systems, which limits the effectiveness of the VLAN groupings. Other limitations of existing VLAN implementations include excessive broadcast traffic (which consume both network bandwidth and end system CPU bandwidth), disallowing transmissions out multiple ports, hop-by-hop switching determinations, and requiring multi-protocol routers to enable transmission between separate VLANs. Another problem with many VLAN switched networks is that although they allow a meshed topology, none of the redundant links can be active at the same time. Generally, the active links are determined by a spanning tree algorithm which finds one loop-free tree-based path through the network. Unfortunately, any links or nodes not in the active tree path are placed in standby.
Thus, there are numerous limitations with many prior switched communications networks.
SUMMARY OF THE INVENTION
In accordance with certain broad aspects of the present invention, methods and apparatus are provided which enable one or more of the following services:
directory (distributed discovery of MAC addresses and protocol alias addresses)
topology (distributed topology protocol exchanges among access and network switches)
broadcast resolution (resolution of broadcast frames to unicast frames at access switches)
policy (applying security restrictions prior to connection setup)
path determination (determine multiple paths from source to destination)
connection management (source-routed mapping of connections on a desired path)
call rerouting (distributed rerouting of a call when a link fails)
broadcast/unknown service (restricted flooding of nonresolvable packets)
connection-oriented switching (source-destination MAC addresses used as a connection key)
According to a first aspect of the invention, a fully distributed switching model is provided in which each switch is capable of processing all aspects of the call processing and switching functionality. Each switch maintains a local directory of locally-attached end systems on access ports. As each local end system generates MAC frames, the switch “learns” the source MAC frame as well as any higher level protocol address information; these higher layer addresses are referred to as alias addresses since they alias (or rename) the MAC end system. Thus, all end system network and MAC mappings are discovered automatically at each access port on the switch.
The local directory may also store local VLAN mappings. VLAN mappings identify the logical LAN to which the switch port or user belongs. A logical or virtual LAN allows users to appear as being on the same physical (or extended) LAN segment even though they may be geographically separated. By default, all ports and users are on a “default” or base VLAN.
More specifically, the VLAN-IDs are used only for policy and to scope broadcast/unknown destinations.
With each access switch having its own locally learned mappings in the directory, there is a “virtual directory” which provides a scalable, demand-based mechanism for distributing directory mappings through the switch domain. The virtual directory is defined as the collective directory mappings that exist in each switch within the domain. So, at all times, the virtual directory always has the complete mappings of all known users within the domain. It is not necessary to distribute or synchronize the switches' directory between themselves. Rather, each switch may access its local directory cache for locally attached end systems, and if the end system is not found in the local directory cache, there is triggered a query to the virtual directory, i.e., to each of the remote switches local directory. This means that at any given access switch, virtual directory queries are made only for destination addresses that are not in the local directory.
The call-originating switch which cannot resolve a mapping locally within its own directory, issues a resolve to the virtual directory by “VLAN ARPing”. This is similar to how IP hosts resolve destination IP addresses to MAC addresses, but instead of “ARPing” to all end systems, the VLAN resolve message is sent only to other switches within the VLAN domain. Only the switch having the local directory mapping of the requested resolve information, known as the “owner” switch, will respond; multiple owners may exist if an end system is redundantly connected. All resolutions are then stored as remote entries in the call-originating switches' remote directory. The owner switch is stored along with the resolve information in the remote directory. The combination of the local directory and inter-switch resolve messaging provides mobility (i.e., end systems can attach anywhere in the network).
The directory of resolved mappings becomes in essence another cache. These entries reflect active or attempted connectivity resolutions, so the cache is self-adjusting for the source-destination traffic. It will size automatically to the actual resolution requirements of the call-originating switch.
Another important aspect of the invention is to provide topology and connection services which include the following:
distributed link state protocol
distributed path determination
distributed connection management
distributed threading the needle
distributed call rerouting.
The topology services are built into every switch, which allows each switch to be completely autonomous in its behavior, yet provides the necessary functionality across the entire switching fabric.
The switches run a di
Andlauer Philip
Bahi David H.
Cullerot David L.
Dobbins Kurt
Fee Brendan
Enterasys Networks Inc.
Hsu Alpus H.
Lahive & Cockfield LLP
Nguyen Toan D.
LandOfFree
Distributed connection-oriented services for switched... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Distributed connection-oriented services for switched..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Distributed connection-oriented services for switched... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3289996