Multiplex communications – Pathfinding or routing – Switching a message which includes an address header
Reexamination Certificate
1999-01-21
2003-05-27
Vincent, David (Department: 2661)
Multiplex communications
Pathfinding or routing
Switching a message which includes an address header
C370S225000, C370S244000
Reexamination Certificate
active
06570881
ABSTRACT:
FIELD OF THE INVENTION
The invention relates generally to network systems, such as local area networks (LANs) with structures and systems providing high availability interfaces. More particularly, the invention relates to packet based high-speed meshes constructed with a set of loosely coupled switches, optionally a configuration protocol, trunk network interfaces and optionally a reachability protocol wherein each switch provides a single shared LAN by interconnecting two or more links.
BACKGROUND OF THE INVENTION
Systems are known which provide redundant capabilities. Redundant packet forwarding devices have been used to handle the possibility of failed links and/or failed packet forwarding devices. Such packet forwarding devices have been used with protocols which protect available paths and are used for rerouting traffic.
Traditionally the spanning tree protocol has been use to provide both Layer 1 and 2 redundancy, in switch networks, by configuring alternate paths. When a link or switch element fails, a backup link is activated. The Spanning Tree algorithm defines the active and backup links in the topology. Spanning Tree is designed to control the many switches of the topology and has a slow (in the tens of seconds) reconfiguration time.
The reconfiguration time of Spanning Tree is a significant disadvantage. As network systems become more complex and handle additional traffic, the reconfiguration time becomes increasingly problematic. Further, spanning tree disables the redundant path. This negates the possibility of using redundant paths for increases in throughput.
Packet based high-speed meshes called trunk clusters offer a significant benefit as to providing redundancy and switch network systems as well as providing load sharing capabilities using multiple switches which provide additional switch throughput for each additional switch provided in the switch cluster or trunk cluster. Multiple switches forming a single logical switch provide layer 1 and layer 2 redundancy and switch network systems which avoid the use of spanning tree. These systems are based on multiple switches forming a single logical switch that can participate in topology control protocols (e.g. spanning tree and Trunk Cluster Management Protocol (TCMP)) and can share MAC address learning information.
U.S. application Ser. No. 09/014,548 filed Jan. 28, 1998, now U.S. Pat. No. 6,195,351 (which is hereby incorporated by reference) provides a Logical Switch Set (LSS) formed of two or more switches that act as a single packet forwarding device with specific connection rules. The LSS may be used as either a redundant switch set (RSS)or as a Load Sharing Switch Set (LSSS). The maximum throughput of the LSSS increases with each additional switch. A LSSS can only interconnect with the other devices via trunked links that contain at least one physical connection to each switch. This prevents any single link attachments to the LSSS. This also implies that if a link is lost, leaving no remaining links to that switch, then the LSSS can either chose to drop service to that station or stop using that switch for load sharing for all connected devices. Effectively removing that switch from the LSSS.
The LSS implements a single logical device, which is comprised of two or more separate switches. This logical device can operate as a simple packet forwarding device, a full function fully manageable switch, or any variation between these extremes. The simplest non filtering packet forwarding device need not share any state information between devices. A full function fully manageable switch model requires that state or parameter changes, acquired from the traffic streams a switch receives, be shared between the other switches. State and parameter changes can occur from learning MAC source addresses and processing management and control frames. Management and control frames destined for the logical device will be directed to one of the switches by the traffic steering algorithm of an attached device. Each switch in the LSS implements one or more of the physical links, within a trunked group, for each of the ports of the logical device. Each switch must have a unified view of the logical switch and port parameters to properly forward traffic and to transmit management and control frames. This sharing of state information requires a communication path between switches in a LSS. This communication path may be a dedicated resource or travel in band.
U.S. application Ser. No. 09/014,547 of Jan. 28, 1998, now U.S. Pat. No. 6,195,349, discloses a grouping of switch elements (Application Ser. No. 09/014,547 of Jan. 28, 1998 is hereby incorporated by reference) which is referred to herein as a trunk cluster. The trunk cluster cooperates with edge devices to provide a scalable logical LAN. The trunk cluster is constructed with a set of loosely coupled switches, a configuration protocol, trunked network interfaces, and optionally a reachability protocol.
Each switch in the trunk cluster provides a single “shared LAN” by interconnecting two or more links. The edge devices attached to the links run a trunk configuration protocol. These attached edge devices aggregate two or more physical ports into a trunked port. The devices view each trunked port as if the attached device is connected to a shared logical LAN with multiple other attached devices.
This logical LAN is designed to provide scaleability and resilience. The set of devices that interconnect to a logical LAN, called Edge devices, cooperate using configuration protocols and traffic steering methods required by the logical LAN. The trunk cluster is comprised of two or more trunk switches. A single logical LAN is provided with the edge devices splitting the traffic (directing traffic flow) across the links in a trunked port. Each trunk switch provides a separate path within the trunk cluster (multiple parallel paths are provided). The two or more separate paths between edge devices allow this logical LAN to increase bandwidth by adding more trunk switches and automatically decrease bandwidth in the event of a link failure and/or in the event of a trunk switch failure.
As each switch only provides a “shared LAN” that carries only part of the traffic between edge devices, each switch does not need to and must not participate in any topology control or discovery protocol. Spanning tree, TCMP, IGMP, and GARP packets are flooded. Unicast MAC (Media Access Controller) source addresses are learned and used to intelligently forward/filter unicast packets to minimize flooding within the “shared LAN” and increase throughput. The maximum throughput of the Trunk Cluster increases with each additional switch.
Each MAC device of an edge device may transmit a hello signal to MAC devices of other edge devices as part of a trunk configuration protocol. Such a configuration protocol configures the connections or link connections between the switches and the edge devices and also preferably continuously verifies the links status. The hello signal includes a trunk or edge device ID identifying the respective edge device of the MAC device transmitting the hello signal. Each MAC device records the edge device ID's of the hello signals received from other edge devices. These recorded edge device ID's are formed into an hello list for each MAC device. The TCMP agent of an edge device forms a trunk list for each other edge device. Each trunk list for a particular other edge device includes MAC addresses of the present edge device which received the hello signals from the respective one of the other edge devices. For example, if edge device A had three MAC devices which received hello signals from edge device B, edge device A would have a trunk list for edge device B which contained those three MAC devices. When edge device A received traffic for edge device B, edge device A would divide the traffic among the three MAC devices in the trunk list for edge device B. This dividing of traffic received by edge device A for edge device B, is according to the standard trunking convention known a
Heiner, Jr. Edward A.
Hiscock James Scott
Wils Joris Johannes Maria
3Com Corporation
McGlew and Tuttle , P.C.
Phunkulh Bob A.
Vincent David
LandOfFree
High-speed trunk cluster reliable load sharing system using... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with High-speed trunk cluster reliable load sharing system using..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and High-speed trunk cluster reliable load sharing system using... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3030766