Electrical computers and digital processing systems: multicomput – Computer-to-computer data routing
Reexamination Certificate
1999-09-17
2004-08-03
Thompson, Marc D. (Department: 2142)
Electrical computers and digital processing systems: multicomput
Computer-to-computer data routing
C709S227000, C370S351000
Reexamination Certificate
active
06772219
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a message relaying scheme that is suitable for constructing a ultra-high-speed message relaying system.
2. Description of the Background Art
In recent years, in conjunction with the explosive spread of the so called Internet, the throughput improvement in a connection-less network for leading the so called IP message to a desired terminal has become a major technical problem.
In the case of connection-less network, the end-to-end communications are provided as each routing processing device refers to a destination address assigned to each individual message and transfers the individual message independently to a desired output path. For this reason, it is possible to utilize the bandwidth of each physical link more flexibly compared with the connection-oriented network, and there is an advantage in that the bandwidth of the physical link can be utilized efficiently in a situation where traffics transmitted by terminals are highly bursty, which is characteristic of applications such as WWW for which the spontaneity in the continuous information communications has low importance.
On the other hand, there is a problem in that the amount of calculations required to the routing processing device at a time of routing individual message is overwhelmingly large compared with the connection-oriented network so that it is difficult to realize a high throughput routing processing device.
This problem stems both from the large amount of calculations that is inherently associated with the connection-less communications due to the fact that a physical link to which each message should be outputted must be determined whenever a message arrives, and from the physical constraint associated with electric circuits that it is very difficult to construct ultra-high-speed signal transmission paths using electric signals. This problem becomes particularly noticeable in the case where the physical link accommodated by the routing processing device uses various protocols such as Ethernet, PPP, ATM and SDH, or in the case of providing the so called multicast connection in which a single message is to be copied and lead to plural output paths.
A conventionally adopted method for resolving this problem is one in which a plurality of routing processing devices are directly connected by a physical link group and operated such that they can be regarded as a single routing processing device as a whole, as shown in FIG.
57
.
In outline, the ultra-high-speed message relaying system shown in
FIG. 57
operates as follows. When a message is entered from a connection-less network
103
-i to a routing processing device
102
-i, the routing processing device
102
-i extracts a destination address of the entered message, obtains a next hop address of that message by referring to a routing table which has the extracted destination address as a key, selects a physical link corresponding to that next hop address from physical links connected to the own device, and transmits the message toward that physical link. When this operation is carried out by every routing processing device
102
-i, the message transfer between connection-less networks
103
-i via a plurality of routing processing devices
102
-i can be realized. At this point, the routing processing at each routing processing device
102
-i is carried out independently from any other routing processing device
102
-j so that the throughput of the routing processing for the system as a whole can be improved.
However, this scheme directly arranges the physical links among the routing processing devices
102
-i so that the bandwidth of the communication path between each connected routing processing devices
102
-i becomes fixed and the transmission bandwidth allocation between the routing processing devices cannot follow the dynamic change of traffics among the connection-less networks. For this reason, this scheme has a problem that it is impossible to utilize the bandwidth of the physical link efficiently.
In particular, in the case of applications such as WWW in which the user's access target changes frequently, messages transmitted from one routing processing device
102
-i are not uniformly distributed over other connection-less networks so that the bandwidth required between the routing processing devices
102
-i and
102
-j changes largely in time, but this scheme cannot provide the required bandwidth between the routing processing devices
102
-i and
102
-j in such a situation. Consequently, this scheme has a problem that the comfortable communications cannot be provided to users because a length of a queue for holding messages to be transmitted to a physical link with insufficient bandwidth that is connected to a connection-less network
103
-i becomes long so that the message delay time becomes extremely long.
As described, the conventional routing processing device has been associated with a problem that it is difficult to realize a high throughput routing processing device because of the large amount of calculations that is inherently associated with the connection-less communications and the physical constraint associated with electric circuits. This problem becomes particularly noticeable in the case where the routing processing device accommodates various protocols or in the case of providing the multicast connection.
In addition, the conventional ultra-high-speed message relaying system formed by combining the routing processing devices, the transmission bandwidth allocation between the routing processing devices is fixed so that the bandwidth cannot be allocated by following the traffic pattern that continually changes in time, and therefore there has been a problem that the comfortable communications cannot be provided to users.
SUMMARY OF THE INVENTION
It is therefore an object of the present invention to provide a message relaying scheme suitable for constructing a ultra-high-speed message relaying system, which is capable of allocating the bandwidth by following the traffic pattern between the routing processing devices that are continually changing in time and transferred by various protocols.
It is another object of the present invention to provide a message relaying scheme capable of providing the multicast connection by resolving the problem due to the large amount of calculations that is inherently associated with the connection-less communications and the physical constraint associated with electric circuits.
According to one aspect of the present invention there is provided a message relay device connected with a plurality of routing processing devices for carrying out connection-less communications, for relaying a message from one routing processing device to another routing processing device, the message relay device comprising: a plurality of interface units provided in correspondence to the routing processing devices; a switching unit connecting the plurality of interface units; and a command unit for commanding the switching unit to set up a bypass communication channel corresponding to an internal identifier which is an identifier defined within the message relay device for identifying at least a target interface unit corresponding to a relaying target routing processing device of each message; wherein each interface unit determines a flow to which an input message entered from a corresponding routing processing device of each interface unit belongs and assigns the internal identifier to the input message according to the flow, transmits the input message to the switching unit, and receives an output message to be outputted to the corresponding routing processing device of each interface unit from the switching unit; and the switching unit sets up the bypass communication channel according to the internal identifier in response to a command from the command unit, and switches messages transmitted from one interface unit to another interface unit using the bypass communication channel.
In this aspect of the present invention, the command
Kabushiki Kaisha Toshiba
Thompson Marc D.
Vu Thong
LandOfFree
Message relaying scheme based on switching in units of flows does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Message relaying scheme based on switching in units of flows, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Message relaying scheme based on switching in units of flows will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3317296