JP2001345849A - Ip priority control method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a service for guaranteeing a rage, equal at least to a contracted band from a network to a user by combining technology for deciding whether the present transmitting rate of the user exceeds the band contracted with the network on a core-stateless network. SOLUTION: Each time an IP packet arrives at an edge router, the transmission rate of the user sending that IP packet is measured, and the measured transmission rate is written in the header of that IP packet. On the basis of the transmission rate written in the header of the IP packet, the packet is selectively abandoned. In such a core-stateless network, the edge router decides whether the present transmission rate of the user exceeds the band contracted with the network.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a third protocol.
The present invention relates to an IP priority control method for performing priority control in an IP layer for use when IP (Internet Protocol) is used for a layer.

[0002]

2. Description of the Related Art As a method of performing priority control in an IP layer, a core stateless network has been proposed (for example, Ion Stoica et al., Core-Stateless F.).
air Queuing: Achieving Approximately Fair Band
width Allocations in High Speed Networks, "SI
GCOMM 98, 1998.).

In a core stateless network, an edge router manages information for each flow in order to realize fair bandwidth allocation, but a core router does not.
In the queuing method for the core stateless network, an arrival rate is measured by an edge router, and the values are written in a header of an IP packet.

[0004] A core stateless network is composed of an edge router and a core router. The required system configuration is shown in FIGS. FIG.
FIG. 1 shows the basic configuration of a conventional general router. In FIG. 1, A denotes a routing unit, and B and C denote FIG.
It corresponds to the functions shown in (a) and (b).

FIG. 2A shows a function required for processing when an IP packet arrives at an edge router in a conventional core stateless network.
(B) shows a function required to determine whether to input an IP packet to a buffer in a conventional core stateless network.

In the edge router, as shown in FIG. 2A, every time an IP packet arrives, the user identification unit 1
Identify from which user the IP packet was sent. Then, the rate information writing unit 2 measures the transmission rate of the user sending the IP packet, and writes the value of the measured transmission rate into the header of the IP packet.

Further, in the edge router and the core router, as shown in FIG. 2B, the header reading unit 3 reads the value of the transmission rate written in the header. Then, based on the information, the selective packet discarding unit 4 determines whether or not to input the IP packet to the buffer 5.

In the core stateless network, the edge router needs to manage information for each user in order to measure a transmission rate for each user, as described above. However, since the core router only uses information given to the header of the IP packet,
There is no need to manage information for each flow. as a result,
It has the feature of being excellent in scalability.

Also, TSW (Time Sliding Window)
Has been proposed to determine whether the current transmission rate of a user exceeds the bandwidth contracted with the network (for example, D. Clark et al, Explici).
t Allocation of Best-Effort Packet Delivery
Service, "IEEE / ACM transaction on networkin
g vol.6 no.4, August, 1998).

In the TSW, three parameters, Win_length, Avg_rate, and T_front, are used as shown in FIG. 3 (for initial values, see FIG. 3I). As described below,
Some of these three parameters are updated each time an IP packet arrives (see FIG. 3T). Then, if Avg_rate is equal to or greater than the rate contracted with the network, it is written that the user exceeds the band contracted with the network at the probability P = (Avg_rate-contracted rate) / Avg_rate. Otherwise, the header of the IP packet is written as not exceeding the bandwidth contracted with the network.

[0011]

However, in the above described core stateless network, it is determined whether or not the current transmission rate of the user exceeds the bandwidth contracted with the network, and this is determined by processing the IP packet. The idea of combining with the technology to be reflected in the past has not been shown before.

An object of the present invention is to solve the above-mentioned problems in the prior art, and to determine whether or not a current transmission rate of a user exceeds a band contracted with the network in a core stateless network. By combining the technologies to be implemented, the network realizes an IP priority control method capable of providing a service that guarantees a rate to a user at least equal to the contracted bandwidth while having scalability. is there.

[0013]

In order to achieve the above object, an IP priority control method according to the present invention provides, whenever an IP packet arrives at an edge router, a transmission rate of a user sending the IP packet. A core stateless which measures and writes the measured transmission rate in the header of the IP packet, and selectively discards the packet at the edge router and the core router based on the transmission rate written in the header of the IP packet. An IP priority control method in a network, wherein the edge router
It is characterized in that it is determined whether or not the current transmission rate of the user exceeds the bandwidth contracted with the network.

[0014] Further, the IP priority control method according to the present invention comprises:
In the above determination, when it is determined that the current transmission rate of the user exceeds the bandwidth contracted with the network, with a certain probability, information that exceeds the contracted bandwidth,
A value obtained by subtracting the contracted bandwidth from the user's current transmission rate is written in the header of the IP packet.

Further, the IP priority control method according to the present invention
In the above determination, the present invention is characterized in that packet discarding is selectively performed only on the IP packet in which the information on the current transmission rate is written in the header based on the value thereof.

In the IP priority control method according to the present invention, when it is determined in the above determination that the current transmission rate of the user does not exceed the bandwidth contracted with the network, information indicating the fact is provided. , Written in the header of the IP packet.

The differences from the prior art are as follows. In the IP priority control method according to the present invention, in the core stateless network, the edge router determines whether or not the rate currently transmitted by the user exceeds the bandwidth contracted with the network. However, in a conventional core stateless network, such a determination has not been made.

Further, in the IP priority control method according to the present invention, information as to whether the current transmission rate exceeds the contracted bandwidth is written in the header of the IP packet. However, in the conventional core stateless network, such information is not written in the header of the IP packet.

Further, in the IP priority control method according to the present invention, the rate information is written into the header of the IP packet at a certain probability when it is determined that the contracted bandwidth is exceeded as a result of the determination. However, in the conventional core stateless network, transmission rate information is written in the header of every IP packet.

Further, in the IP priority control method according to the present invention, when reading the header of an IP packet, information on whether or not the transmission rate currently used for transmission exceeds the contracted band and rate information are read. read. However, in the conventional core stateless network, only the rate information is read. It is characterized by the following.

The features of the IP priority control method according to the present invention are as follows.
In a core stateless network, the edge router determines whether or not the rate that the user is currently transmitting exceeds the bandwidth contracted with the network, and consequently determines that the rate exceeds the bandwidth contracted with the network. Only when is the rate information written in the header of the IP packet.

[0022]

Embodiments of the present invention will be described below in detail with reference to preferred embodiments shown in the drawings.

FIGS. 4 and 5 show the system configuration of the embodiment. Here, FIGS. 4 and 5 correspond to FIGS. 2A and 2B, respectively, showing the prior art, and FIG. 6 is a diagram showing the state of an IP packet in a specific example. It is.

FIG. 4 shows processing functions required when an IP packet arrives at an edge router in the IP priority control method according to the present invention.
, The user identification unit 1 determines from which user the arriving IP packet is transmitted. The contract bandwidth determination unit 6 determines that the current transmission rate of this user is
It is determined whether the user has exceeded the bandwidth contracted with the network.

In FIG. 4, the contract bandwidth determination unit 6
Transmits the result of the above determination to the rate information writing unit 2. The rate information writing unit 2 writes the following information into the header of the IP packet based on the received determination result.

For example, if it is determined that the transmission rate of the user exceeds the bandwidth contracted by the user with the network, a certain probability P (in the case of using the TSW shown in FIG. 3, P = ( Avg_rate-contracted rate) / Avg_
In the header, a value obtained by subtracting the contracted band from the value “1” and the current transmission rate value (when using the TSW shown in FIG. 3 (Avg_rate−contracted band)) Is written).

In all other cases, only the value "0" is written to all headers. Here, a value “1” means that the transmission rate of the user exceeds the bandwidth contracted by the user with the network, and a value “0” indicates that the transmission rate of the user exceeds the bandwidth contracted by the network with the user. Means not exceeding.

FIG. 5 shows a processing function required to determine whether or not to input an IP packet to a buffer in the IP priority control method according to the present invention. The unit 3 reads the information written in the header by the rate information writing unit 2 in FIG.

Here, if it is read that the value "0" is written in the header, the IP packet does not pass through the selective packet discarding unit 4 and is directly input to the buffer. When it is read that the value “1” is written in the header, the value of the rate information written in the header is transmitted to the selective packet discarding unit 4.

The selective packet discarding unit 4 determines from the received transmission rate value whether to selectively discard the IP packet. If it is determined that selective packet discarding is performed, the IP packet is discarded at that point. If it is determined that selective packet discarding is not to be performed, the IP packet is directly input to the buffer 5.

In the example shown in FIG. 6, two IP packets having an IP packet size of 0.5 M bits arrive at the edge router at time 1.0 seconds and time 1.2 seconds, respectively. Here, the contract bandwidth between the user and the network is 1M (bit
/ s), it is assumed that the edge router uses the TSW for the contract bandwidth determination unit 5 and the rate information writing unit 2, and the description is continued. Here, the initial values of the parameters of the TSW are, as shown in FIG. 3I, respectively Win_lengh =
1 (second), Avg_rate = 1M (bit / s)), T_front = 0
(Seconds).

In the system set as described above,
When the first IP packet arrives, the contract bandwidth determination unit 6 updates the TSW parameters. As a result, Avg_rate = 0.75M (bit / s), T_front = 1
(Seconds). The details are as follows. Byte_in_TSW = 1M (bit / s) × 1 (second) = 1M (bi
t) New_Bytes = 1M (bit) + 0.5M (bit) = 1.5M
(Bit) Avg_rate = 1.5M (bit) / (1.0-0 + 1) (sec)
= 0.75M (bit / s) T_front = 1 (second)

Here, the calculated Avg_rate (= 0.7)
5M (bit / s)) is a value smaller than 1M (bit / s), which is the contracted bandwidth between the user and the network. Therefore, the contracted bandwidth determination unit 6 determines that the current user transmission rate is equal to or less than the contracted bandwidth. Judgment is made and the result is transmitted to the rate information writing unit 2. The rate information writing unit 2 receives the information that the current user's transmission rate is equal to or less than the contracted band from the contracted band determining unit 6, and writes the value “0” into the header of the first IP packet.

When it is determined whether or not this first IP packet is input to the buffer 5, first, the header reading unit 3 reads information written in the header. Here, since the value “0” is written in the first IP packet, the value is directly input to the buffer without passing through the selective packet discarding unit 4.

When the second IP packet arrives, the contract bandwidth determination unit 5 updates the TSW parameters. As a result, Avg_rate = 1.04M (bit / s), T_
front = 1.2 (seconds). The details are as follows. Byte_in_TSW = 0.75M (bit / s) × 1 (second) = 0.
75M (bit) New_Bytes = 0.75M (bit) + 0.5M (bit) = 1.
25M (bit) Avg_rate = 1.25M (bit) / (1.2-1.0 + 1)
(Second) = 1.04M (bit / s) T_front = 1.2 (second)

Here, the calculated Avg_rate is greater than 1M (bit / s), which is the contracted bandwidth between the user and the network. Therefore, it is determined that the current user transmission rate is equal to or greater than the contracted bandwidth, and the result is determined. Rate information writing unit 2
Send to Since the rate information writing unit 2 has received the information that the bandwidth is equal to or more than the contract bandwidth, the rate information writing unit 2 determines to write the rate information with a certain probability P.

Here, the description is continued on the assumption that it is determined that the rate information is to be written. Rate information writing unit 2
Is a value obtained by subtracting the contracted bandwidth from the current transmission rate (1.04M-
(1.0M) = 0.04M (bit / s) is written. When determining whether or not the second IP packet is input to the buffer, first, the header reading unit 3 reads information written in the header.

Since the value “1” is written in the second IP packet, the header reading unit 3 sends the value of the rate information written in the header to the selective packet discarding unit 4. The selective packet discarding unit 4 determines whether or not to selectively discard the IP packet based on the received transmission rate value. In this case, it is determined that the IP packet is discarded. Discarded at That is, the second IP packet is not input to the buffer 5.

Here, the above-mentioned probability will be described. The above-mentioned probability P is used when the contract bandwidth determination unit 6 in FIG. 6 determines whether or not the current transmission rate of a certain user exceeds the bandwidth contracted by the user with the network. is there. This probability P is expressed as
As mentioned above, if Avg_rate> contracted rate,
P = (Avg_rate-contracted rate) / Avg_rate, where 0 <P <1.

The selective packet discarding unit 4 in FIG.
However, the probability B used when judging whether to discard the IP packet transmitted with the value “1” written in the header portion of the IP packet from the header reading unit 3 is given by target queue length / current When the queue length is a parameter m, it can be defined as follows.

(Equation 1)

Here, m> 1 means that the current queue length is smaller than the target queue length. On the other hand, m
≦ 1 means that the current queue length is greater than or equal to the target queue length. When m is smaller than 1, the IP packet is more actively discarded than when m is larger than 1. L is a packet label value, and α is a fair rate value calculated by an algorithm described later. Note that the initial value of α is the link band, and k is a constant (see FIG. 7).

When m is greater than 1 and L is between mα and α, the larger the value of L, the higher the discard probability.
Since the IP packet whose value of L is close to mα is discarded with a high probability, the flow rate responding to the congestion is usually equal to or lower than mα, and is expected to be stabilized. Therefore, a flow whose rate exceeds mα can be regarded as a flow that does not respond to congestion. An IP packet whose L value is mα is always discarded.

For an IP packet determined not to be discarded by the above-described stochastic packet discarding algorithm, the router inputs the IP packet to the buffer as long as the buffer has a free space. FIG. 8 shows the code of the above-described stochastic packet discard algorithm.

FIG. 9 shows an algorithm for calculating the fair rate value α described above. The router selects an IP packet determined to be input to the buffer with the probability Pa, and reads the label value of the IP packet. If the number of such packets becomes Na, the average of those values is
The fair rate value α is obtained.

The result of computer simulation of the above-described IP priority control method will be briefly described below. FIG.
Shows a two-node model of 10 flows. Congestion occurs on the link between Router 1 and Router 2. Router 1 is the bottleneck point. The transmission delay of each link was set to 5 ms.

In the first simulation, all flows were Reno TCP. RenoTCP has fast retransmit and fast recovery in addition to slow start and congestion avoidance. FIG. 11 shows the ideal band allocation value and the measured band allocation value.
The ideal bandwidth allocation value and the measured bandwidth allocation value are almost the same. This has confirmed the effectiveness of the IP priority control method according to the present invention.

In the second simulation, at the start of the simulation, the values in Table 1 shown in FIG. However, after 30 seconds, the values of all the minimum bands were set to 1M (bit / s). FIG.
Fig. 3 shows an ideal bandwidth allocation value from 2 to 30 seconds and a measured bandwidth allocation value. FIG. 14 shows the ideal bandwidth allocation values from 30 seconds to 60 seconds and the measured bandwidth allocation values. In both cases, the ideal bandwidth allocation and the measured bandwidth allocation are almost the same. This has confirmed the effectiveness of the IP priority control method according to the present invention.

In addition to the above, performance evaluation and the like in environments with different transmission delays were also performed, and the effectiveness of the IP priority control method according to the present invention was also confirmed here.

The above embodiment is merely an example of the present invention, and the present invention is not limited to the embodiment.

[0050]

As described in detail above, according to the present invention, the bandwidth contracted to the network is provided to each user while having the characteristic of the core stateless network having excellent scalability. Has a remarkable effect that a service capable of guaranteeing the minimum can be realized.

[Brief description of the drawings]

FIG. 1 is a diagram showing a basic configuration of a conventional general router.

FIG. 2 (a) shows a function required when an IP packet arrives at an edge router in a conventional core stateless network, and FIG. 2 (b) inputs an IP packet to a buffer in a conventional core stateless network. FIG. 3 is a configuration diagram of functions necessary for determining whether or not the function is performed.

FIG. 3 is a diagram illustrating an algorithm of TSW.

FIG. 4 is a configuration diagram of functions required when an IP packet arrives at an edge router in the embodiment.

FIG. 5 is a configuration diagram of functions required when determining whether to input an IP packet to a buffer in the embodiment.

FIG. 6 is an explanatory diagram of an IP packet in a specific example.

FIG. 7 is a diagram illustrating a relationship between a value of L and a probability B;

FIG. 8 is a diagram showing codes of a stochastic packet discarding algorithm.

FIG. 9 is a diagram showing an algorithm for calculating a fair rate value α.

FIG. 10 is a diagram showing a two-node model of 10 flows used in the simulation.

FIG. 11 is a diagram (part 1) showing an ideal band allocation value and a measured band allocation value, which are simulation results.

FIG. 12 is a diagram showing a table used for a simulation.

FIG. 13 is a diagram (part 2) showing an ideal band allocation value and a measured band allocation value, which are simulation results.

FIG. 14 is a diagram (part 3) showing an ideal band allocation value and a measured band allocation value, which are simulation results.

[Explanation of symbols]

 Reference Signs List 1 User identification unit 2 Rate information writing unit 3 Header reading unit 4 Selective packet discarding unit 5 Buffer 6 Contract bandwidth judgment unit

Continuation of the front page (72) Inventor Ikuo Yamazaki 2-3-1 Otemachi, Chiyoda-ku, Tokyo F-term (reference) in Nippon Telegraph and Telephone Corporation 5K030 HA08 HB13 HC01 HD03 LA03 LC09 LC15 LE05 MA04 MB09 5K033 CB06 CB17 DA05 DB16 DB18 EA07 EC04

Claims (4)

[Claims] Each time an IP packet arrives at an edge router, a transmission rate of a user sending the IP packet is measured, and the measured transmission rate is written in a header of the IP packet. An IP priority control method in a core stateless network, wherein a packet is selectively discarded based on a transmission rate written in a header of the IP packet, wherein the edge router uses a current transmission rate of a user. A determination is made as to whether the bandwidth exceeds the bandwidth contracted with the network. 2. In the above determination, when it is determined that the current transmission rate of the user exceeds the bandwidth contracted with the network, there is a certain probability that information indicating that the current transmission rate exceeds the contracted bandwidth, 2. The IP priority control method according to claim 1, wherein a value obtained by subtracting the contracted bandwidth from the transmission rate is written in the header of the IP packet. 3. The method according to claim 2, wherein the information of the current transmission rate is selectively discarded only for the IP packet written in the header based on the value thereof. IP priority control method. 4. When the current transmission rate of the user is determined not to exceed the bandwidth contracted with the network in the determination, information indicating the fact is written in the header of the IP packet. 3. The IP priority control method according to claim 2, wherein