Multiplex communications – Pathfinding or routing – Switching a message which includes an address header
Reexamination Certificate
1997-10-29
2001-02-06
Pham, Chi H. (Department: 2731)
Multiplex communications
Pathfinding or routing
Switching a message which includes an address header
C370S395430, C370S396000
Reexamination Certificate
active
06185213
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a node device (including a router device for relaying packets) for transferring packets from one logical network to another logical network, and a packet transfer control method to be carried out at a node device.
2. Description of the Background Art
In the packet communication network for transferring packets on a network layer such as that of the Internet protocol, logical networks are defined. Within the same logical network, the packet transfer is carried out in a prescribed frame format according to a protocol (Ethernet, point-to-point link, ATM, frame relay, etc.) of a datalink layer constituting that logical network.
On the other hand, in a case of transferring a packet to the different logical network, an address information of the packet is analyzed at a router located on a boundary between logical networks, and the packet is relayed by being transferred toward a prescribed logical network. Consequently, a packer relay transfer processing based on the address analysis at a router will be carried out as many times as a number of logical network boundaries that the packet passes through.
As the datalink network constituting individual logical network becomes faster, there arises a problem that such a relay transfer processing based on the address analysis at a router becomes a bottleneck for the overall transfer performance. In order to resolve this problem, there has been a proposition of a method for carrying out a cut-through transfer in which the address analysis processing for each packet at a router is not required so as to significantly improve a router processing performance, as described in detail in the IETF RFC 1953 and 1954.
At a router that carries out this cut-through transfer scheme, the information such as an address and an upper level protocol is analyzed for the first one or few packets, and prescribed control messages are exchanged with a neighboring router (a previous or next hop router belonging to the identical logical network) on a route of the analyzed packet flow so as to store the information on the analyzed packet flow by mapping it to a header value (a VPI/VCI value of a cell header in a case of ATM) that can be processed at high speed.
Then, the transfer processing for subsequent packets belonging to that packet flow are carried out according to the stored header value that can be processed at high speed. By carrying out the above processing at each router on the route of the packet flow, it becomes possible to carry out the fast transfer (referred to as a cut-through transfer hereafter) processing for packets other than the first few packets. Also, by notifying a request on the communication quality from an end-host, it also becomes possible to provide a cut-through path that satisfies a requested communication quality with respect to an end-to-end flow.
The above described packet transfer control method sets up a cut-through path for each communication between specific transmitting host and receiving host, for example, so that a number of cut-through paths to be managed by the router can be potentially large. Also, there is a need to carry out the address analysis processing for each packet just as in the conventional router until the cut-through path is established, so that there can be a router at which the address analysis processing for each packet may cause the bottleneck.
For this reason, there is also a proposition of a method in which a cut-through path between specific routers is set up in advance at a time of the network activation, for example, instead of setting up a dedicated cut-through path for a specific end-to-end packet flow at a time of detecting that packet flow. In this case, it is impractical to set up a cut-through path dedicatedly with respect to every conceivable end-to-end packet flow, so that a cut-through path will be set up to have a general availability (such that all packet flows destined to a specific logical network from one router can use it commonly, for example). Consequently, various end-to-end packet flows are going to share the cut-through path formed from one router to another router of another logical network.
In this method, it is possible to eliminate a possibility for causing the bottleneck due to the conventional address analysis processing until the cut-through path is formed as encountered in the earlier described method. However, in this method, when a scale of the network is increased, a number of cut-through paths to be set up in advance also increases, so that it is preferable to limit a number of cut-through paths to a reasonable level (such that the cut-through path is set up in advance only for those packet flows which are destined to a logical network which is known to have a large traffic, for example). In such a case, when some end-to-end packet flow occurs, if there is no cut-through path set up in advance on the route of that packet flow, the conventional transfer processing based on the address analysis for each packet must be carried out at routers on that route, and therefore the problem of the performance bottleneck arises again.
Also, in this method for setting up a cut-through path with a certain degree of general availability, it is impossible to provide a dedicated cut-through path that satisfies the requested communication quality with respect to a specific end-to-end packet flow.
SUMMARY OF THE INVENTION
It is therefore an object of the present invention to provide a packet transfer control method and a node device which are capable of providing a dedicated cut-through path for a specific end-to-end packet flow, while reducing a time required until the cut-through path is established, so as to be able to reduce the load of the address analysis processing at the router.
According to one aspect of the present invention there is provided a method for transferring packets from one node belonging to one logical network, comprising the steps of: storing at said one node an information indicating an existence of a first dedicated virtual path for transferring packet flows specified by a first condition, which is capable of transferring packets without applying a network layer processing up to another node belonging to another logical network different from said one logical network; and communicating an identifier information for identifying a second dedicated virtual path for transferring a packet flow specified by a second condition which is more specific than the first condition by utilizing the first dedicated virtual path, and a condition information for indicating the second condition, in order to set up the second dedicated virtual path up to a termination point node of the first dedicated virtual path or a further downstream side node, between said one node and the termination point node.
A datalink layer of a network to which the present invention applies can be any of the Ethernet, point-to-point link, ATM, frame relay, etc. Also, a network layer protocol can be any of the IP, IPX, etc.
According to the present invention, the first dedicated virtual path (cut-through path) is set up in advance for packet flows specified by the first condition, so that packets belonging to these packet flows can be transferred from the own node to another node which is not a neighboring node of the own node, without applying the network layer processing at intermediate nodes between them.
In a case of ATM, for example, this first dedicated virtual path can be a cut-through formed by a VP (Virtual Path, in which a plurality of VCs (Virtual connections) are contained) or a cut-through formed by a plurality of PVCs (Permanent Virtual Connections). Also, in a case of Ethernet, for example, this first dedicated virtual path can be a cut-through realized by transferring a MAC frame with a value indicating the first dedicated virtual path registered in a specific region, and routing the MAC frame by referring or rewriting this specific region with a higher priority over another specific region to be describe
Katsube Yasuhiro
Nagami Kenichi
Tanaka Hisako
Foley & Lardner
Kabushiki Kaisha Toshiba
Pham Brenda H.
Pham Chi H.
LandOfFree
Packet transfer control method and node device 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 Packet transfer control method and node device using..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Packet transfer control method and node device using... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2595966