JP2003023456A - Multiplexer, band controller, program and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a band controller, which by being installed in a communication network and impartially distributes an extra band between connections in a normal state (in a non-congested state) while warranting a minimum contracted band, even in a congestion state for each connection established in the communication network, so as to more efficiently utilize network resource, economically realizes efficient utilization, even in a high-speed channel, and is installed in a node so as to realize a simple hardware configuration, thereby realizing congestion avoidance in a node, with high efficiency. SOLUTION: A maximum queue length is preset for a buffer, a continuous function of queue length X which becomes zero, when it exceeds the maximum queue length is referred to as α (X), and the sum of the weights of connections i at which one packet or more is stored in the buffer is defined as Wact. Then a permitted queue length AQ is calculated as AQ=α (X).Wi/Wact, and packets belonging to connections i, whose queue length Xi by each connection i is the permissible queue length AQ or below are discriminated to be acceptable.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a device for guaranteeing the minimum bandwidth of each connection accommodated in a communication network.

[0002]

2. Description of the Related Art In recent years, data traffic such as IP traffic has increased exponentially due to the rapid spread of the Internet and LAN. As a result, the frequency of congestion on the network has increased, and the deterioration of service quality for users has become a problem. For example, although the best-effort service that does not guarantee the transfer quality has been mainstream in the conventional Internet, the number of cases in which a sufficient throughput cannot be obtained only by the best-effort service is increasing. It is considered that there will be an increasing demand for services that guarantee the minimum bandwidth and delay quality for each line when connecting with a high-speed line.

There is a GFR (Guaranteed Frame Rate) service as a service that guarantees the minimum bandwidth contracted for each user. At the time of non-congestion, each user can share and use the available bandwidth (remaining bandwidth).

Many proposals have been made regarding a method for realizing a GFR service, one of which is WRR (Wei
ghted Round Robin) method. FIG. 15 is a diagram showing an outline of the WRR system. The WRR method has a queue for each connection and a weight is assigned to each connection.
Bandwidth is distributed among connections by performing read control according to the weight for each queue.

By accommodating all the connections sharing the bandwidth in the WRR, the bandwidth can be distributed according to the weight of the accommodated connection. With the WRR method, it is possible to evenly distribute the excess band between the connections according to a certain rule while guaranteeing the minimum band.

However, the complexity of the WRR system hardware is proportional to the number of connections to be accommodated, and becomes a bottleneck for speeding up. In addition, since it is necessary to have a queue for each connection, the hardware amount of the buffer section becomes a problem. That is, the WRR system becomes faster, and when the number of accommodated connections increases, it is difficult to economically realize the GFR service. Therefore, a method for realizing the GFR service with a simple hardware configuration is required.

A FIFO-Tagging method is known as a method for realizing a GFR service with a simple hardware configuration. In this method, it is not necessary to have a buffer for each connection, but only one buffer for each line. FI
In the FO-Tagging method, the input rate to the network is observed at the entrance of the network for each connection, and if the measured rate is less than the MCR (Minimum Cell Rate), the packet of that connection passes through and exceeds the MCR. If so, a tag (hereinafter, Tag) is added to the header part of the packet. Here, the MCR is a band in which the network guarantees the transfer for the connection.

The FIFO buffer is provided with a threshold value and constantly monitors whether or not the queue length exceeds the threshold value. If the queue length exceeds the set threshold value, the packet with Tag in the header is FI.
Only packets without Tag attached to the FIFO buffer, that is, packets with a connection whose input rate to the network is less than or equal to MCR, are passed before being entered into the FO buffer.

In the FIFO-Tagging system, FI
MC of the connection that accommodates the output speed of the FO buffer
By setting the total R or more, it is possible to guarantee the MCR at the time of congestion. Also, when there is no congestion, M of each connection
If there is a band exceeding the sum of CRs and a surplus band, the band is shared by multiple connections.

[0010]

However, since the buffer is shared by a plurality of connections in this method, the band exceeding the sum of the MCRs of each connection and the surplus band are proportional to the input rate to the FIFO buffer. "Fairness" because it is distributed in the form of each connection
There was a problem in that respect. In the FIFO-Tagging method, there is no regulation on how to distribute the surplus bandwidth between connections when a surplus bandwidth occurs, so in an extreme case, one connection occupies the surplus bandwidth, which is an unfair situation. There was a problem that.

The present invention has been made against such a background, and by installing it in a communication network, each connection accommodated in the communication network can be contracted even at the time of congestion. While guaranteeing the bandwidth, during normal times (when there is no congestion), the excess bandwidth is evenly distributed among the connections.
An object of the present invention is to provide a multiplexing device, a band control device, a program, and a recording medium which can enable more efficient use of network resources and can be economically realized even in a high-speed line. Furthermore, by being installed in a node such as a packet switching switch, a simple hardware configuration can be realized, and efficient congestion avoidance in the node, a multiplexing device, a bandwidth control device, a program, and a recording medium can be realized. The purpose is to provide.

[0012]

The first aspect of the present invention is to:
Means for identifying the connections 1 to k to which the arriving packet belongs, and means for determining whether or not the packet can be accepted,
This determining means is a multiplexing device provided with a means for multiplexing the packets that have been admitted.

Here, the features of the present invention are that the means for detecting the queue length Xi of the multiplexing means for each connection i (i is any of 1 to k) and the predetermined for each connection i. Means for holding the value of the assigned weight Wi, and a maximum queue length that is a threshold value is set in advance in the multiplexing means, and the sum ΣXi of the queue lengths Xi that becomes substantially zero when the maximum queue length is exceeded. Let α (ΣXi) be the continuous function of 1
The sum of the weights of the connection i in which the above packets are accumulated is set to Wact, and the determining unit calculates the admission queue length AQi as AQi = α (ΣXi) · Wi / Wact and a unit for each connection i. Queue length X
It is provided with a unit for determining that a packet belonging to a connection in which i is less than or equal to the permission queue length AQi is admitted.

Thus, the multiplexer of the present invention is installed in the communication network to ensure the minimum bandwidth of each connection, while considering the MCR contracting the excess bandwidth and the usage status of the network resources. Then, since the connection is distributed among the connections, the communication network can be effectively used. Further, since the present invention adopts the FIFO having a simple hardware configuration, it can be economically realized even in a high-speed line.
Furthermore, by installing in a node such as a packet switching switch, a simple hardware configuration can be realized and efficient congestion avoidance in the node can be realized.

The means for calculating the ratio of the queue length Xj of the connection j (j is one of 1 to k) rejected by the judging means to the permitted queue length AQj and the calculating means. The packet of the connection j determined to be rejected by the determining unit with a probability according to the determined ratio is regarded as a packet determined to be acceptable, and is stored in the multiplexing unit. You can also

As a result, as compared with the case where the packet of the connection exceeding the permitted queue length is suddenly rejected, the slow traffic control can be performed, and the stable traffic control can be realized.

The determining means may include means for determining a packet to be rejected when the packet of the connection j to be rejected is not present in the multiplexing means.

That is, the fact that the packet of connection j does not exist in the multiplexing means means that the packet of connection j is sporadic. Even if such sporadic packets are rejected and discarded, the effect of avoiding congestion is extremely low as compared with the case of discarding continuous packets. Therefore, it is desirable not to discard such sporadic packets meaninglessly.

A packet arriving at the network that exceeds the minimum guaranteed bandwidth is tagged, and the judging means judges that the packet is to be rejected when the tag is to be rejected. It is desirable to include means for determining a packet to be accepted as acceptance. As a result, a packet transfer service that guarantees the minimum bandwidth can be realized.

The means for detecting the queue length Xi may be configured to include means for detecting the queue length Xi for a change cycle of the queue length Xi which is less than a predetermined change cycle, or the means for detecting the queue length Xi may be It is also possible to adopt a configuration that includes means for outputting the maximum value of the queue length Xi within the unit time for each unit time as the detection result of the queue length Xi. As a result, it is possible to absorb an instantaneous change in the queue length Xi and avoid a malfunction of traffic control due to disturbance.

The multiplexing means includes means for transmitting a monitoring packet, and the means for detecting the queue length Xi is input to the multiplexing means by the monitoring packet transmitted from the means for transmitting the monitoring packet. It is also possible to include means for detecting the queue length Xi according to the time difference between the time and the time at which the monitoring packet is output from the means for multiplexing.

As a result, for example, the effective queue length Xi including parameters such as the actual delay time can be detected as compared with the detection of the queue length Xi based only on the number of accumulated packets.

A plurality of service classes are provided, means for identifying the service class to which the arriving packet belongs, means for detecting the queue length Xi for each service class of the multiplexing means for each connection i, and connection means for holding a value of the weight Wi for each service class that is predetermined for each i, and the determining means is the permission queue length AQi for each service class.
For AQi = α (ΣXi) · Wi / Wact, and for packets belonging to a connection whose queue length Xi for each service class for each connection i is less than or equal to the permissible queue length AQi for each service class. It is also possible to provide means for determining acceptance.

For example, the packets belonging to one connection are further classified into service classes of which the real-time property is important and other packets, and the packet of which the real-time property is important is weighted more than the other packets. If the value is increased, the packet for which real-time property is important can be read out smoothly, so that the QoS (Quality of Service) for the user can be improved.

Further, the judgment means comprises means for calculating a ratio of the queue length Xj of the connection j (one of 1 to k) rejected by the judgment means to exceed the permitted queue length AQj. It is also possible to provide means for rejecting at least two consecutive packets of the connection j according to the calculation result of the calculating means.

It is also possible to provide means for prohibiting successive acceptance refusals for packets of connection j whose acceptance is denied according to the calculation result of the calculating means.

Further, the connection j
It is also possible to provide a means for switching and selecting between the means for rejecting acceptance of the packet and the means for inhibiting the continuous rejection of acceptance.

For example, when the queue length Xj suddenly exceeds the permitted queue length AQj, it is necessary to perform rapid traffic control in order to avoid the occurrence of congestion. When such a situation occurs, it can be dealt with by continuously discarding packets. On the other hand, when the queue length Xj gently exceeds the permitted queue length AQj, it is possible to perform gentle traffic control and realize stable traffic control. It is desirable to switch and use such contradictory control modes depending on the situation. Such switching control is preferably provided as an additional function of the determining means.

Alternatively, a means for accumulating in the multiplexing means after delaying a fixed time according to the ratio calculated by the calculating means to the packet of the connection j judged to be rejected by the judging means It can be provided.

As a result, the RTT (Round Trip Time) is increased without discarding the packet, and the effect of lowering the transmission rate for the packet transmission source can be expected, so compared to the case where the packet is discarded. Traffic can be used effectively.

Alternatively, the connection identification numbers are classified into a plurality of groups in advance, and the determining means is
For a packet whose connection identification number of a packet to be determined to be rejected to be accepted belongs to the above-mentioned group, it is possible to include means for determining a packet to be determined to be rejected to be acceptable. As a result, it is possible to perform gradual traffic control and realize stable traffic control.

A second aspect of the present invention is a band control device including the multiplexing device of the present invention. By providing the multiplexing device of the present invention in the bandwidth control device, for each connection accommodated in the communication network, while guaranteeing the minimum bandwidth contracted even at the time of congestion, at the normal time (at the time of non-congestion) Can distribute the surplus bandwidth fairly among the connections and enable more efficient use of network resources, and can also realize an economically feasible bandwidth control device even for high-speed lines. .

A third aspect of the present invention is a program characterized by causing the information processing device to realize the function of the multiplexing device of the present invention by being installed in the information processing device. Alternatively, the program is characterized by causing the information processing device to realize the function of the bandwidth control device of the present invention by being installed in the information processing device.

A fourth aspect of the present invention is a recording medium readable by the information processing device, in which the program of the present invention is recorded.

[0035]

1 is a block diagram of a multiplexer according to an embodiment of the present invention.
And it demonstrates with reference to FIG. FIG. 1 is a block configuration diagram of a multiplexer according to an embodiment of the present invention. FIG. 12 is a diagram for explaining queue length detection using a monitoring packet.

According to the present invention, as shown in FIG. 1, a connection identification unit 1 for identifying the connections 1 to k to which an arriving packet belongs, an acceptance determination unit 5 for determining acceptance or rejection of the packet, and this acceptance determination unit. 5 is a multiplexing device including a FIFO type buffer 6 for temporarily storing packets according to the determination result of 5.

Here, the feature of the present invention is that the queue length Xi of the buffer 6 for each connection i is calculated from among the packets of a plurality of connections accumulated in the buffer 6.
Queue length monitoring unit 7 and virtual queue management unit 3 for detecting
And a weight Wi predetermined for each connection i
And the connection information storage unit 2 that holds the value of Qma.
x is set and a continuous function of the sum ΣXi of the queue lengths Xi that becomes almost zero when the maximum queue length Qmax is exceeded is defined as α (ΣXi), and one or more packets are accumulated in the buffer 6 for the connection i. The sum of weights is Wac
and the permission queue length calculation unit 4 determines that the permission queue length AQi
Is calculated as AQi = α (XΣi) · Wi / Wact, and the acceptance determination unit 5 determines that the packet belonging to the connection whose queue length Xi for each connection i is equal to or less than the permission queue length AQi is permitted. It is in.

In this embodiment, the queue length monitoring unit 7 and the virtual queue length management unit 3 connect to the actual queue length of the buffer 6 under the condition that a plurality of connections are mixed and accumulated in one buffer 6. The detection of the queue length Xi of the buffer 6 for each i is called a virtual queue length. In the following, for simplification of description, the virtual queue length is simply referred to as the queue length Xi or Xj.

The admission judgment unit 5 calculates the ratio of the queue length Xj of the connection j rejected to be accepted to the admission queue length AQj, and the connection j judged to be rejected with a probability corresponding to the calculated ratio. It is also possible to regard a part of the packet as a packet determined to be admitted and store it in the buffer 6.

Alternatively, the acceptance determination unit 5 determines that the queue length Xj of the connection j whose acceptance is rejected is the permitted queue length AQ.
The ratio exceeding j is calculated, and according to this calculation result, two consecutive packets of the connection j are rejected or rejected, or continuous rejection of rejection is prohibited. The user can arbitrarily select the switching of the acceptance refusal policy. For example, when the calculated ratio exceeds a predetermined value, at least two packets of the connection j are continuously rejected, and when the calculated ratio is less than the predetermined value, the rejected connection is rejected. It is also possible to prohibit successive rejections of j packets.

Further, the acceptance judging unit 5 can judge that the packet which should be judged to be rejected is accepted when the packet of the connection j which should be judged to be rejected is not present in the buffer 6.

Packets arriving at the network that exceed the minimum guaranteed bandwidth are tagged, and the admission judgment unit 5 judges that the packet is to be rejected when the tag is added to the packet. A packet that should be accepted is determined to be admitted.

The acceptance judgment unit 5 calculates the ratio of the queue length Xj of the connection j whose acceptance is rejected exceeds the permitted queue length AQj, and the connection j which is determined to be acceptance rejected.
It is possible to store the packet in the buffer 6 after delaying the packet for a certain time according to the calculated ratio.

The queue length monitoring unit 7 detects the queue length Xi for a change period of a predetermined change period of the queue length Xi or less. Alternatively, the queue length monitoring unit 7 outputs the maximum value of the queue length X within a unit time as the detection result of the queue length Xi for each unit time.

Further, as shown in FIG. 12, the buffer 6 is provided with a monitoring packet inserting section 8 for transmitting a monitoring packet, and the queue length monitoring section 7 receives the monitoring packet sent from the monitoring packet inserting section 8 from the buffer 6. It is also possible to detect the queue length Xi according to the time difference between the time input to the buffer and the time when the monitoring packet is output from the buffer 6.

Further, a plurality of service classes are provided,
The connection identification unit 1 simultaneously identifies the connection of the arriving packet and the service class to which the packet belongs, and the virtual queue management unit 3 detects the queue length Xi of each buffer class of the buffer 6 for each connection i, The connection information storage unit 2 holds the value of the weight Wi predetermined for each service class for each connection i, and the acceptance determination unit 5 sets the permission queue length AQi for each service class to AQi = α (ΣXi). Wi / Wact is calculated, and it is also possible to judge that the packet belonging to the connection in which the queue length Xi for each service class for each connection i is the permissible queue length AQi for each service class or less is admitted.

Further, the connection identification numbers are classified into a plurality of groups in advance, and the admission judgment unit 5 judges that the packets whose connection identification numbers of the packets to be judged to be refusal of acceptance belong to the predetermined group are rejected. It is also possible to judge that the packet to be judged is admission permitted. As an example of classification of the groups, the connection identification numbers can be classified into even numbers or odd numbers.

The multiplexing device of the present invention can be used as a band control device (UPC: Usage Parameter Control) by being provided in a network as shown in FIG. Further, by providing the switch in the packet switching switch, it can be used for avoiding the congestion of the buffer in the switch.

The multiplexing device or the band control device of the present invention is installed in a computer device which is an information processing device, and a program for causing the computer device to realize the function of the multiplexing device or the band control device of the present invention. It can be realized by installing in a computer device. This program is recorded on a recording medium,
The computer device installs the program of the present invention by reading the recorded contents of the recording medium,
Alternatively, the program of the present invention is directly installed from a server holding the program of the present invention via a network.

The embodiments of the present invention will be described in more detail below.

FIG. 14 is a block diagram of a network to which the present invention is applied to provide a minimum bandwidth guarantee service. The originating user sends a packet through the network to the terminating user. The network guarantees the transfer rate up to the minimum bandwidth contracted to the user. A plurality of users are accommodated in each link of the network, and the surplus bandwidth is distributed among the users according to a weight determined based on a charge or a minimum guaranteed bandwidth.

First, when the originating user sends a packet to the network, the rate observing device located at the ingress edge of the network measures the packet sending rate of the user to the network. If the packet transmission rate exceeds the minimum guaranteed bandwidth, Tag is attached to the packet header, and packets transmitted below the minimum guaranteed bandwidth are directly transmitted to the network. The minimum bandwidth is guaranteed by transferring the packet without Tag attached to the called user without discarding it in the network. The multiplexer of the present invention is located in a relay node and performs a process of distributing a link band shared by a plurality of users to each user according to a weight.

(First Embodiment) A first embodiment of the present invention will be described with reference to FIGS. As shown in FIG. 1, the multiplexing device of the present invention comprises a connection identification unit 1, a band control unit 10, and a buffer unit 11. The connection identification unit 1 identifies the connection from the header of the arrived packet. The bandwidth control unit 10 performs processing relating to bandwidth control based on the connection information. The buffer unit 11 is a simple FIFO buffer 6 and is shared by all users accommodated in the link.

A queue length monitoring unit 7 is connected to the buffer 6, and transmits the queue length of the buffer 6 and information regarding input / output packets at the time of packet input / output to the virtual queue management unit 3 in the preceding stage. Here, the information regarding the input / output packet includes the packet length of the packet input to or output from the buffer 6 and the identification number of the connection to which the packet belongs.

When a packet arrives from the input line side, the connection identification unit 1 identifies the connection of the packet, sends it to the bandwidth control unit 10 in the subsequent stage, determines whether or not to input it to the buffer 6, and discards it. The determined packet is discarded on the spot, and the packet that is not determined to be discarded is input to the buffer unit 11 in the subsequent stage and output to the output line according to the FIFO standard.

In the present invention, band control based on packet discard in the band control unit 10 plays an important role. Therefore, the processing of the bandwidth control unit 10 will be described in detail. The band control unit 10 includes a virtual queue management unit 3, a permission queue length calculation unit 4, a connection information storage unit 2, and an acceptance determination unit 5.

The virtual queue management unit 3 calculates the queue length of the buffer 6 and the queue length of each connection based on the information from the queue length monitoring unit 7. The virtual queue management unit 3 is connected to the connection information storage unit 2, and the calculated virtual queue length for each connection is recorded in the connection information storage unit 2.

A method of calculating each queue length will be described.
First, regarding the calculation of the queue length of the buffer 6, the queue length information transmitted from the queue length monitoring unit 7 is simply retained in the virtual queue management unit 3. Next, regarding the method of calculating the virtual queue length for each connection, since the buffer 6 is empty in the initial state, the virtual queue length of each connection is zero. When a packet is input to the buffer 6, the packet length of the packet and the connection identification number are transmitted through the queue length monitoring unit 7. Based on the information, the virtual queue management unit 3 accesses the connection identification unit 1 and adds the transmitted packet length to the virtual queue length of the relevant connection. At the time of outputting a packet, on the contrary, the process of subtracting the transmitted packet length from the virtual queue length is performed.

FIG. 2 shows an example of the connection information storage unit 2 in which information about the virtual queue length for each connection calculated by the virtual queue management unit 3 and information about the connection weight are stored.

Next, the permission queue length calculation unit 4 receives information on the queue length of the buffer 6 from the virtual queue management unit 3 and the connection information storage unit 2 and information on the virtual queue length and weight for each connection and sends them. It has a function to calculate the permission queue length. Here, the connection permitted queue length represents the maximum queue length permitted to be input to the buffer 6.

A method of calculating the admission queue length AQi of the connection i will be specifically described. When the connection weight is Wi and the current queue length of the buffer 6 is Xi, AQi = α (ΣXi) · Wi / Wact is calculated. α (ΣXi) is a continuous function, and the graphs in FIGS. 3 to 9 are examples of the queue length function α (ΣXi). The horizontal axis represents the queue length Xi, and the vertical axis represents the function α (Σ
Take the value of Xi). Wact is the sum of the weights of active connections. Here, the active connection is a connection whose virtual queue length is greater than zero, that is, a connection in which at least one packet is stored in the buffer 6.

The function α (ΣXi) becomes zero when the queue length exceeds a certain threshold (Qmax in the figure). That is, when the queue length exceeds a certain threshold, it is determined that there is congestion, and the permission queue length of each connection becomes zero.

For example, in the example of FIG. 3, the function α (ΣXi) takes a constant value in the range not exceeding the threshold Qmax, and the function α (ΣXi) becomes zero when the threshold Qmax is exceeded. That is, the permitted queue length AQi is distributed in proportion to the connection weight within the range not exceeding the threshold value Qmax.

In the example of FIG. 4, the function α (ΣXi) gradually decreases from the zero value of the queue length Xi to a predetermined value less than the threshold value Qmax, and suddenly decreases as it approaches the threshold value Qmax. , The threshold value Qmax is zero. That is, when the queue length Xi is small, the permission queue length AQi is distributed more than the weight of the connection i, but when the queue length Xi exceeds a predetermined value, the permission queue length AQi sharply increases.
The distribution rate of Qi decreases.

In the example of FIG. 5, the queue length Xi and the function α (Σ
The value of Xi) is linearly inversely proportional. That is, the permitted queue length AQi is linearly inversely proportional to the queue length Xi. This is the simplest and most basic control example.

In the example of FIG. 6, the queue length Xi and the function α (Σ
The value of Xi) is inversely proportional to a quadratic function. That is, the permitted queue length AQi is large when the queue length Xi is small, but starts to decrease suddenly as the queue length Xi increases, and decreases gradually as the queue length Xi approaches the threshold value Qmax. As a result, it is possible to strongly suppress the increase in traffic when the queue length Xi starts to increase.

In the example of FIG. 7, the queue length Xi and the function α (Σ
The value of Xi) is inversely proportional to a quadratic function that is the inverse of the quadratic function shown in FIG. That is, the permission queue length A
Qi is large when the queue length Xi is small, but begins to decrease gradually as the queue length Xi increases.
The decrease becomes steeper as i approaches the threshold value Qmax. This makes it possible to strongly suppress the increase in traffic as the queue length Xi approaches the threshold value Qmax.

In the example of FIG. 8, the queue length Xi and the function α (Σ
The value of Xi) is inversely proportional to the value of stepwise. That is, the permitted queue length AQi is large when the queue length Xi is small, but gradually decreases as the queue length Xi increases,
It suddenly starts to decrease from the predetermined queue length Xi, and decreases gradually again as the queue length Xi approaches the threshold value Qmax. As a result, the increase in traffic can be strongly suppressed when the queue length Xi is near the middle of zero and the threshold value Qmax.

In the example of FIG. 9, the queue length Xi and the function α (Σ
The value of Xi) is inversely proportional to the value of Xi) in a stepwise manner, contrary to the example of FIG. That is, the permitted queue length AQi is large when the queue length Xi is small, but starts to decrease suddenly as the queue length Xi increases, starts to decrease gradually from the predetermined queue length Xi, and decreases rapidly again as the queue length Xi approaches the threshold value Qmax. Begins to decrease. As a result, the increase in traffic can be strongly suppressed at the time when the queue length Xi starts to increase from zero and when it approaches the threshold value Qimax.

Although the current queue length (instantaneous value of the queue length when a packet arrives) is used as the queue length Xi, an average value of the queue length within a fixed time may be adopted as the queue length Xi. .

The admission judgment unit 5 stores the packets arriving from the permission queue length and the virtual queue length for each connection in the buffer 6
Input to or discard. FIG. 10 is a flowchart showing a processing flow upon arrival of a packet for performing definite discard processing. As shown in FIG. 10, if the virtual queue length Xi is equal to or less than the permitted queue length AQi, the buffer 6
If it exceeds the permitted queue length AQi, Tag
Packets marked with are discarded, and packets not tagged are input to the buffer 6,
The minimum bandwidth of each connection is always guaranteed, and control is performed so that the occupancy of the buffer 6 becomes fair.

If the virtual queue length exceeds the permitted queue length, it is possible to stochastically discard the packets instead of deterministically discarding the packets as described above. FIG. 11 is a flowchart showing a processing flow upon arrival of a packet for performing probabilistic discard processing. As shown in FIG. 11, the permitted queue length is AQi and the virtual queue length is X.
Given i, the discard probability P is given by P = (Xi-AQi) / Xi. In other words, when the virtual queue length exceeds the permitted queue length, it is dropped by the excess rate. When the bandwidth is controlled by probabilistic discard, if one connection is focused, the phenomenon that packets are continuously discarded even when the virtual queue length exceeds the permitted queue length is deterministic packet discard. It can be said that the affinity with the rate control of TCP becomes higher because the number of cells is smaller than that of the case of TCP. The packet input to the buffer 6 is a FIFO
It is output to the output line according to the standard.

The pseudo random number Rand in FIG. 11 is generated every packet arrival, and the discard probability P and the pseudo random number R are generated.
By executing the discarding based on the result of comparison with and, it is possible to actually realize the discarding with the discard probability P. For example, when the discard probability P is 0.5, the probability that the pseudo random number Rand takes a value larger than 0.5 is also 0.5, and when the pseudo random number Rand is larger than 0.5, discarding is performed. By executing the execution, the disposal is performed according to the disposal probability P.

The multiplexer of the present invention estimates the congestion state of the network based on the queue length, and if it is judged that there is congestion, the permitted queue length becomes equal to zero, and the network is transmitted to the network at a rate below the minimum guaranteed bandwidth. The network forwards only outgoing packets. When the queue length is less than or equal to a threshold, it is determined that there is a surplus bandwidth, and each connection occupies a buffer proportional to the weight, and the bandwidth is evenly distributed among the connections.

Further, according to the above-described embodiment, while the GFR service is realized by the FIFO buffer having the simple hardware configuration, the packet read control is carried out by individually holding the buffer for each connection such as the WRR method. As with the method, it is possible to distribute the excess bandwidth fairly among the connections. In addition, since one buffer is shared by a large number of connections, there is an advantage that the required buffer amount can be reduced due to the statistical multiplexing effect, as compared with a system having a separate buffer for each connection.

(Second Embodiment) In the second embodiment of the present invention, as a function of the acceptance judgment unit 5 shown in FIG. 1, the queue length Xj of the connection j rejected by the acceptance judgment unit 5 exceeds the permitted queue length AQj. Is calculated, and when the calculated ratio exceeds a predetermined value, at least two consecutive packets of connection j are rejected, and when the calculated ratio is less than or equal to the predetermined value, reception is accepted. Prohibit successive rejections of rejected packets of connection j.

That is, when the queue length Xj suddenly exceeds the permitted queue length AQj, it is necessary to perform abrupt traffic control in order to avoid the occurrence of congestion. When such a situation occurs, it can be dealt with by continuously discarding packets. On the other hand, when the queue length Xj gently exceeds the permitted queue length AQj, it is possible to perform gentle traffic control and realize stable traffic control.

(Third Embodiment) In the third embodiment of the present invention, as a function of the acceptance judgment unit 5 shown in FIG. 1, when the packet of the connection j to be judged to be rejected is not present in the buffer 6, it is judged to be rejected. A packet that should be accepted is determined to be admitted.

That is, the fact that the packet of connection j does not exist in the buffer 6 indicates that the packet of connection j is sporadic. Even if such sporadic packets are rejected and discarded, the effect of avoiding congestion is extremely low as compared with the case of discarding continuous packets. Therefore, such sporadic packets should not be meaninglessly discarded.

(Fourth Embodiment) The fourth embodiment of the present invention has a function of the admission judgment unit 5 shown in FIG. To do. This fixed time is the queue length Xj
Varies depending on the ratio at which the permission queue length AQj is exceeded.

As a result, the RTT (Round Trip Time) becomes large even without discarding the packet, and the effect of lowering the transmission rate to the packet transmission source can be expected. Therefore, as compared with the case of discarding the packet. Traffic can be used effectively.

(Fifth Embodiment) In the fifth embodiment of the present invention, the function of the admission judgment unit 5 shown in FIG. 1 is to classify the connection identification numbers into a plurality of groups in advance, and When the connection identification number belongs to a predetermined group, it can be determined that the packet is admitted. For example, with respect to either the even numbered packet or the odd numbered packet, the packet which should be determined to be rejected is determined to be accepted. As a result, it is possible to perform gradual traffic control and realize stable traffic control.

(Sixth Embodiment) In the sixth embodiment of the present invention, as a function of the queue length monitoring unit 7 shown in FIG. 1, the queue length Xi is detected for a change period of a predetermined change period or less of the queue length Xi. As a result, it is possible to absorb an instantaneous change in the queue length Xi and avoid a malfunction of traffic control due to disturbance.

(Seventh Embodiment) In the seventh embodiment of the present invention, the maximum value of the queue length Xi within the unit time is used as the detection result of the queue length Xi as the function of the queue length monitoring unit 7 shown in FIG. Output every time. As a result, similarly to the sixth embodiment, it is possible to absorb an instantaneous change in the queue length Xi and avoid a malfunction in traffic control due to a disturbance.

(Eighth Embodiment) The eighth embodiment of the present invention is shown in FIG.
Will be described with reference to. As an eighth embodiment, as shown in FIG. 12, a buffer 6 is provided with a monitoring packet insertion unit 8 for sending a monitoring packet, and the queue length monitoring unit 7 is arranged so that the monitoring packet sent from the monitoring packet insertion unit 8 is The queue length X is detected according to the time difference between the time input to the buffer 6 and the time when the monitoring packet is output from the buffer 6.

As a result, the effective queue length Xi including parameters such as the actual delay time can be detected as compared with the detection of the queue length Xi based on the number of accumulated packets.

(Ninth Embodiment) The ninth embodiment of the present invention is shown in FIG.
Will be described with reference to. The ninth embodiment is an example in which packets belonging to one connection are further classified into two service classes and handled. One service class is a real-time (R) class for packets that attach importance to real-timeness, and the other service class is
It is a non-real-time (NR) class for packets that do not place importance on real-time property.

In order to realize the ninth embodiment, the connection identification unit 1 shown in FIG. 1 identifies the connection of the arriving packet and also identifies the service class of the packet. Further, the virtual queue management unit 3 acquires the service class information as well as the connection information of the packets accumulated in the buffer 6. As a result, the connection information storage unit 2 can be provided with a table as shown in FIG.

For example, regarding connection # 1, X belonging to the real time class as the virtual queue length
1 (R) and X1 belonging to the non-real-time class
(NR) is recorded. Further, as the weight, a real-time class weight W1 (R) and a non-real-time class weight W1 (NR) are recorded. Weight W1
(R) is set to be heavier than W1 (NR), and packets belonging to the real-time class can be read with priority over packets belonging to the non-real-time class.

By this means, it is possible to reduce the delay of the packet which attaches importance to the real-time property, so that it is possible to improve the QoS for the user.

[0091]

As described above, according to the present invention,
By installing in the communication network, while guaranteeing the minimum bandwidth contracted for each connection accommodated in the communication network even during congestion, the excess bandwidth is connected during normal time (non-congestion). They can be evenly distributed among the users, and more efficient use of network resources can be enabled. Further, it is possible to realize a multiplexing device and a band control device that can be economically realized even in a high-speed line. Furthermore, by installing in a node such as a packet switching switch, a simple hardware configuration can be realized and efficient congestion avoidance in the node can be realized.

[Brief description of drawings]

FIG. 1 is a block configuration diagram of a multiplexer according to an embodiment of the present invention.

FIG. 2 is a diagram showing an example of a table of a connection information storage unit.

FIG. 3 is a diagram showing an example of a function α (ΣXi).

FIG. 4 is a diagram showing an example of a function α (ΣXi).

FIG. 5 is a diagram showing an example of a function α (ΣXi).

FIG. 6 is a diagram showing an example of a function α (ΣXi).

FIG. 7 is a diagram showing an example of a function α (ΣXi).

FIG. 8 is a diagram showing an example of a function α (ΣXi).

FIG. 9 is a diagram showing an example of a function α (ΣXi).

FIG. 10 is a flowchart showing a processing flow upon arrival of a packet for performing definite discard processing.

FIG. 11 is a flowchart showing a processing flow upon arrival of a packet for performing probabilistic discard processing.

FIG. 12 is a diagram for explaining queue length detection using a monitoring packet.

FIG. 13 is a diagram showing an example of a table of a connection information storage unit of the ninth embodiment.

FIG. 14 is a diagram showing an example of a network configuration to which the present invention is applied.

FIG. 15 is a diagram for explaining an outline of a conventional WRR system.

[Explanation of symbols]

1 Connection identification part 2 Connection information storage 3 Virtual queue management unit 4 Allowed queue length calculator 5 Reception judgment section 6 buffers 7 Queue length monitoring unit 8 Monitoring packet insertion part 10 Band control unit 11 buffer section

Continued front page    (72) Inventor Michihiro Aoki             2-3-1, Otemachi, Chiyoda-ku, Tokyo             Inside Telegraph and Telephone Corporation (72) Inventor Naoaki Yamanaka             2-3-1, Otemachi, Chiyoda-ku, Tokyo             Inside Telegraph and Telephone Corporation F-term (reference) 5K030 GA13 HA08 JA01 KX13 LB19                       LC01 LC11

Claims (17)

[Claims] 1. Multiplexing comprising means for identifying connections 1 to k to which an arriving packet belongs, means for determining acceptance / rejection of the packet, and means for multiplexing packets admitted by the determining means. And a means for detecting a queue length Xi of the multiplexing means for each connection i (i is any of 1 to k), and a means for holding a value of a weight Wi predetermined for each connection i. A maximum queue length serving as a threshold is set in advance in the multiplexing means, and a continuous function of the sum ΣXi of the queue lengths Xi that becomes almost zero when the maximum queue length is exceeded is α
(ΣXi), and the sum of the weights of the connection i in which one or more packets are stored in the multiplexing means is W
ACT, the determining means calculates the permission queue length AQi as AQi = α (ΣXi) · Wi / Wact, and the queue length Xi for each connection i is the permission queue length A
A multiplexing device comprising: means for determining that a packet belonging to a connection of Qi or less is admitted. 2. A means for calculating a ratio of a queue length Xj of a connection j (j is one of 1 to k) rejected by the judging means to exceed the permitted queue length AQj, and a means for calculating the ratio. Connection j determined by the determination means to be rejected with a probability according to the ratio calculated by
2. The multiplexing device according to claim 1, further comprising means for acknowledging a part of the packet as a packet determined to be admitted and accumulating in the multiplexing means. 3. The determining means includes means for determining, when a packet of connection j to be rejected for acceptance is not present in the multiplexing means, a packet to be rejected for acceptance is permitted. 1. The multiplexing device according to 1. 4. A tag is added to a packet arriving at the network that exceeds the minimum guaranteed bandwidth, and the determining unit determines that the packet is to be rejected when the tag is attached to the packet. The multiplexing device according to claim 1, further comprising a unit that determines that a packet that should be determined to be admission is permitted. 5. The multiplexing apparatus according to claim 1, wherein the means for detecting the queue length Xi includes means for detecting the queue length Xi for a change period of a predetermined change period or less of the queue length Xi. 6. The multiplexing according to claim 1, wherein the means for detecting the queue length Xi includes means for outputting the maximum value of the queue length Xi within a unit time for each unit time as a detection result of the queue length Xi. apparatus. 7. The means for sending a monitoring packet to the means for multiplexing, the means for detecting the queue length Xi is the means for multiplexing the monitoring packet sent from the means for sending the monitoring packet. 2. The multiplexing device according to claim 1, further comprising means for detecting the queue length Xi according to a time difference between the input time and the time when the monitoring packet is output from the multiplexing means. 8. A plurality of service classes are provided, means for identifying the service class to which an arriving packet belongs, and means for detecting the queue length Xi for each service class of the multiplexing means for each connection i , A means for holding a value of the weight Wi predetermined for each service class for each connection i, and the determining means determines the permitted queue length AQi for each service class as AQi = α (ΣXi) · Wi / Wact, and a queue length X for each service class for each connection i
The multiplexing device according to claim 1, further comprising: a unit for determining that a packet belonging to a connection having a permissible queue length AQi or less for each service class is admitted. 9. The determining means comprises means for calculating a ratio of a queue length Xj of a connection j (one of 1 to k) refused to be accepted by the determining means to exceed the permitted queue length AQj. The multiplexing device according to claim 1, further comprising means for rejecting at least two packets of the connection j in succession according to the calculation result of the calculating means. 10. The multiplexing apparatus according to claim 9, further comprising means for prohibiting continuous refusal of acceptance for packets of connection j whose acceptance is rejected according to the calculation result of said calculating means. 11. The multiplexing device according to claim 9, further comprising a unit for switching and selecting between a unit for rejecting and accepting packets of the connection j and a unit for inhibiting the successive rejection of packets. 12. The determining means comprises means for calculating a ratio of a queue length Xj of a connection j (any one of 1 to k) refused to be accepted by the determining means, the ratio being greater than the permitted queue length AQj. Connection j judged to be rejected
2. The multiplexer according to claim 1, further comprising means for accumulating in said multiplexing means after delaying said packet by a fixed time in accordance with the ratio calculated by said calculating means. 13. A connection identification number is classified into a plurality of groups in advance, and the determining unit determines that a packet whose connection identification number of a packet to be determined to be rejected belongs to the predetermined group is rejected. The multiplexing device according to claim 1, further comprising means for determining that the packet to be accepted is admitted. 14. A band control device comprising the multiplexing device according to any one of claims 1 to 13. 15. A program which, when installed in an information processing device, causes the information processing device to realize the function of the multiplexing device according to any one of claims 1 to 13. 16. A program, which is installed in an information processing device to cause the information processing device to implement the function of the bandwidth control device according to claim 14. 17. A recording medium readable by the information processing device, in which the program according to claim 15 or 16 is recorded.