JP2003318965A - Packet transfer control method and apparatus, and system apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a packet transfer control method and apparatus, and a system apparatus capable of impartially assigning bands to each IP flow on an IP network. <P>SOLUTION: The packet transfer control apparatus 1 includes: a hop count information learning/storage section 7 for presumedly calculating the hop count value of a packet between two hosts being a generating subjects of the IP flow on the basis of an initial TTL value derived from mount system dependent information acquired by a header analysis section 6 and an observed TTL value acquired by the header analysis section 6 and storing the hop count value for each corresponding IP flow; a transfer priority decision section 8 for selectively deciding the packet transfer priority defined to lower the ranking as the IP flow has a smaller hop count value on the basis of the stored hop count; and a packet transmission control section 9 for performing the packet transfer scheduling based on the decided packet transfer priority. <P>COPYRIGHT: (C)2004,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a packet transfer control method, a device and a system device, and more particularly to I
A packet transfer control method for controlling transfer of packets forming an IP flow on a main backbone line of a P network (IP: Internet Protocol), and a packet transfer control device and a packet transfer control system used directly for the implementation Involved in the device.

[0002]

2. Description of the Related Art In recent years, IP represented by the Internet
With the rapid spread of networks, there is an increasing demand for improving the communication quality of user IP flows in the IP networks.

In order to meet such a demand, the inventors of the present application have studied various factors that deteriorate the communication quality of an IP flow, and as a result, in an IP network, a small number of hop count values between hosts are relatively small. Burst IP flows dominate the line occupancy, and it is not possible to allocate bandwidth fairly to each IP flow existing on the IP network, and this property correlates with deterioration of traffic quality of the entire IP network. It has been experimentally clarified that there is such a result, and the results have already been announced at the Institute of Electronics, Information and Communication Engineers (Technical disclosure document: "Tatsuya Mori, Ryoichi Kawahara, Shozo Naito, Per-
Network traffic analysis focusing on time-block flow statistics, IEICE Technical Report, IN, March 20
02 ”).

On the other hand, IETF (Internet Engineering T
ask force) classifies traffic in an IP network, and QoS (Qual) is classified for each class.
The standardization of the technology (called "Diffserv") that realizes "ity of service" is in progress. This technique classifies traffic at the entrance of the IP network, while performing priority control for each class inside the IP network, thereby attempting to realize the required QoS. .

Therefore, if a small number of dominant IP flows are allocated to a class having a low priority by using the above-mentioned technique, more fair bandwidth allocation is performed for each IP flow on the IP network. It is predicted that it will be possible.

[0006]

However, when the above classification is simply applied to a communication environment in which a large number of unspecified users, such as the backbone line of the Internet, are in fluid use, a small number of dominant However, there is a problem in that it is difficult to quickly identify such an IP flow in the communication process, and as the number of users and the line bandwidth increase, the calculation cost required for classification increases.

Here, the main objects to be solved by the present invention are as follows.

That is, a first object of the present invention is to provide a packet transfer control method, an apparatus and a system apparatus capable of imparting fair bandwidth allocation to each IP flow on an IP network. is there.

A second object of the present invention is to provide a packet transfer control method, a device and a system device capable of performing efficient packet transfer control for each IP flow.

A third object of the present invention is to provide a packet transfer control method, a device and a system device which are extremely easy to set when applied.

Other objects of the present invention will become apparent from the description, drawings, and particularly the description of each claim.

[0012]

First, in the method of the present invention, on the backbone line of an IP network, the header information of a packet forming an IP flow generated in the backbone line is analyzed, and based on the header information, The hop count value of a packet between two hosts, which are the main constituents of an IP flow, is estimated and calculated, the hop count value is accumulated for each corresponding IP flow, and the hop count value is calculated based on the hop count value. A characteristic configuration method is adopted in which packet transfer scheduling is performed according to the packet transfer priority by selectively determining the packet transfer priority defined so that the IP flow having a smaller value has a lower rank.

On the other hand, in the device of the present invention, in an arbitrary area on the backbone line of the IP network,
The header information of the packet forming the IP flow generated in the backbone line is analyzed, and based on the header information, arbitrary mounting system dependent information and observed TTL value (TTL: Time)
To Live. (Surviving time information), and from the acquired mounting system dependent information, the initial TTL value of the corresponding packet is deduced from the header analysis means.
Based on the initial TTL value related to this derivation and the acquired observed TTL value, the hop count value of the packet between the two hosts, which are the main constituents of the IP flow, is estimated and calculated, and the hop count value is corresponded. Based on the hop count information learning / accumulating means for accumulating each IP flow with learning over time, and based on the accumulated hop count value, an IP flow having a smaller hop count value has a lower rank. Characteristic configuration means for providing a transfer priority determining means for selectively determining the defined packet transfer priority and a packet transfer control means for performing packet transfer scheduling according to the determined packet transfer priority Take action.

In the system device of the present invention, I
The header information of the packet forming the IP flow generated in the backbone circuit is analyzed at a plurality of boundary nodes at the connection point between the backbone circuit of the P network and a plurality of peripheral circuits, and the implementation information depends on the arbitrary mounting system from the header information. From the header analysis unit that acquires the information and the observed TTL value and the acquired implementation system-dependent information, the initial TTL value of the corresponding packet is presumedly derived, and
Based on the initial TTL value related to this derivation and the acquired observed TTL value, the hop count value of the packet between the two hosts that are the main components of the IP flow is presumedly calculated, and the hop count value related to this calculation is calculated. , Each corresponding I
Based on the hop count information learning / accumulating means for accumulating with learning over time for each P flow and the accumulated hop count value, the hop count value is small.
A transfer priority determining means for selectively determining a packet transfer priority defined such that the order of P flows becomes lower, and priority information indicating the determined packet transfer priority is provided for each corresponding IP flow. And priority information tagging means for tagging each packet, and a plurality of internal nodes that integrate a plurality of boundary nodes on the backbone line, based on the priority information tagged to the packets at the plurality of boundary nodes, A characteristic configuration means is provided in which a packet transfer control means for performing packet transfer scheduling according to a corresponding packet transfer priority is provided.

More specifically, in order to solve the above problem, the present invention achieves the above-mentioned object by adopting a novel characteristic construction method or means ranging from a superordinate concept to a subordinate concept enumerated below. To be done.

That is, the first feature of the method of the present invention is that, on the backbone line of the IP network, the header information of the packet forming the IP flow generated in the backbone line is analyzed, and the analyzed header information is added to the analyzed header information. On the basis of the above, the hop count value of the packet between the two hosts, which are the main constituents of the IP flow, is estimated and calculated, and the calculated hop count value is accumulated for each corresponding IP flow. Based on the determined hop count value, a packet according to the packet transfer priority is selectively determined by selectively determining a packet transfer priority defined such that the IP flow having a smaller hop count value has a lower rank. This is in adopting a configuration of a packet transfer control method that is performed transfer scheduling.

A second feature of the method of the present invention is that the calculation of the hop count value in the first feature of the method of the present invention is presumedly derived from arbitrary mounting system dependent information that can be obtained from the header information. Initial T of the packet
The packet transfer control method is adopted based on the TL value and the observed TTL value of the packet directly obtained from the header information.

A third feature of the method of the present invention is that the calculation of the hop count value in the second feature of the method of the present invention uses a predetermined calculation formula in which both the initial TTL value and the observed TTL value are variables. This is in adopting the configuration of a packet transfer control method that is performed approximately by using the packet transfer control method.

A fourth feature of the method of the present invention is that the hop count value in the first, second or third feature of the method of the present invention is updated each time a period set to be adjustable is reached. This is due to the adoption of the configuration of the packet transfer control method.

A fifth characteristic of the method of the present invention is that the packet transfer scheduling in the first, second, third or fourth characteristic of the method of the present invention is adjustable with respect to the entire distribution of the hop count value. The packet transfer priority is determined by using two or more queues that are divided by one or more thresholds set in the above, and the packet for each IP flow is assigned to any one of the two or more queues to which the packet corresponds. This is a configuration adoption of a packet transfer control method that is performed stepwise so as to be distributed to

The sixth feature of the method of the present invention is the analysis of the header information, the calculation, accumulation and accumulation of the hop count value in the first, second, third, fourth or fifth feature of the method of the present invention. The packet transfer control method is configured such that the update, the determination of the packet transfer priority, and the packet transfer scheduling are collectively performed in an arbitrary area on the backbone line.

The seventh feature of the method of the present invention is the analysis of the header information, the calculation, accumulation and accumulation of the hop count value in the first, second, third, fourth or fifth feature of the method of the present invention. The updating and the determination of the packet transfer priority are performed in a distributed manner at a plurality of boundary nodes at a connection point between the backbone circuit and a plurality of peripheral circuits, and the packet transfer scheduling is performed by the plurality of boundary nodes at the backbone time. A packet transfer control method, which is a distributed method of a plurality of internal nodes integrated on a line, is adopted.

An eighth feature of the method of the present invention is that the determination of the packet transfer priority at the plurality of boundary nodes in the seventh feature of the method of the present invention provides priority information indicating the packet transfer priority, The packet transfer scheduling in the plurality of internal nodes is performed with the process of tagging the packet for each corresponding IP flow, and the packet transfer scheduling in the plurality of internal nodes is based on the priority information tagged in the packet in the plurality of boundary nodes. This is done by adopting the configuration of the packet transfer control method that is performed.

A ninth characteristic of the method of the present invention is that the hop count value in the seventh or eighth characteristic of the method of the present invention is shared by all of the plurality of boundary nodes at any time. The configuration is adopted.

On the other hand, the first feature of the device of the present invention is a packet transfer control device installed in an arbitrary area on a backbone line of an IP network, and a packet forming an IP flow generated on the backbone line. Of the mounting system dependence information acquired by this header analysis means, and corresponding header analysis means for analyzing any mounting system dependence information and observed TTL value from the header information. While presumably deriving the initial TTL value of the packet, based on the initial TTL value related to this derivation and the observed TTL value acquired by the header analysis means, between the two hosts that are the main constituents of the IP flow Predictively calculates the hop count value of the packet in Based on the hop count information learning / accumulating means for accumulating with time-dependent learning for each corresponding IP flow, and the hop count value concerned based on the hop count value accumulated in the hop count information learning / accumulating means. A transfer priority determining means for selectively determining a packet transfer priority defined such that the IP flow having a smaller value has a lower rank, and the packet transfer priority determined by the transfer priority determining means. A packet transfer control device having a packet transfer control means for performing packet transfer scheduling is employed.

A second feature of the device of the present invention is that the hop count information learning / training in the first feature of the device of the present invention is performed.
A storage unit presets a system-specific initial TTL database for deriving the initial TTL value from the mounting system-dependent information acquired by the header analysis unit, and the initial TTL derived from the system-specific initial TTL database. A TTL database for registering a value and the observed TTL value acquired by the header analysis means in association with each corresponding host IP address, and the initial TT registered in this TTL database
The hop count value approximately calculated using a predetermined calculation formula in which both the L value and the observed TTL value are variables,
The packet transfer control device has a configuration including a flow database for registering in association with each of the corresponding IP flows.

The third feature of the device of the present invention is that the hop count information learning and learning in the second feature of the device of the present invention is performed.
The packet transfer control device is configured so that the accumulating unit has a functional unit that updates the contents of the TTL database and the flow database each time the adjustable period is reached.

A fourth feature of the device of the present invention is that the transfer priority determining means in the first, second or third feature of the device of the present invention is configured such that the packet for each IP flow has the hop count. A function for stepwise determining the required packet transfer priority so that the packet transfer priority can be distributed to any of two or more corresponding queues that are divided by one or more thresholds that are set to be adjustable for the entire distribution of values. Equipped with means,
In the packet transfer control device, the packet transfer control means includes functional means for performing the required packet transfer scheduling using the two or more queues.

Further, the first feature of the system device of the present invention is to integrate a plurality of boundary nodes at a connection point between a backbone line of an IP network and a plurality of peripheral lines and the plurality of boundary nodes on the backbone line. In the packet transfer control system device distributedly arranged with a plurality of internal nodes, the plurality of boundary nodes analyze header information of packets forming an IP flow generated in the backbone line, and extract the header information from the header information. , And a header analysis unit that acquires arbitrary mounting system-dependent information and an observed TTL value, and presumably derives an initial TTL value of the corresponding packet from the mounting system-dependent information acquired by the header analyzing unit. , The observed TTL obtained by the header analysis means and the initial TTL value related to this derivation On the basis of the above, the hop count value of the packet between the two hosts, which are the main constituents of the IP flow, is estimated and calculated, and the hop count value related to this calculation is calculated for each corresponding IP flow over time. Based on the hop count information learning / accumulating means that accumulates with learning, and the hop count value accumulated in the hop count information learning / accumulating means, the IP flow having a smaller hop count value has a lower rank. The transfer priority determining means for selectively determining the packet transfer priority defined as follows, and the priority information representing the packet transfer priority determined by the transfer priority determining means, corresponding to each of the IP flows. And a priority information tagging unit for tagging each of the packets for each of the plurality of internal nodes. A packet transfer control means for performing packet transfer scheduling according to the corresponding packet transfer priority based on the priority information tagged to the packet by the priority information tagging means in the field node, In the adoption of the configuration of the transfer control system device.

A second feature of the system device of the present invention is that the hop count information learning / accumulating means at the plurality of boundary nodes in the first feature of the system device of the present invention is obtained by the header analyzing means. A system-specific initial TTL database preset for deriving the initial TTL value from mounting system-dependent information, the initial TTL value derived from the system-specific initial TTL database, and the observation TTL acquired by the header analysis means And a TTL database for registering the value in association with each corresponding host IP address, and approximation using a predetermined calculation formula in which the initial TTL value and the observed TTL value registered in this TTL database are both variables The calculated hop count value to the corresponding IP flow And a flow database for registering the hop count information so that the hop count information registered in the flow database is shared by all of the plurality of boundary nodes at any time. The packet transfer control system apparatus is configured to include functional means for synchronizing the contents of the TTL database and the flow database in the plurality of boundary nodes with each other.

A third feature of the system device of the present invention is that the hop count information learning / accumulating means at the plurality of boundary nodes in the second feature of the system device of the present invention has a period that is set to be adjustable. In each case, the packet transfer control system device has a configuration for updating the contents of the TTL database and the flow database.

A fourth feature of the system device of the present invention is that the transfer priority determining means in the plurality of boundary nodes in the first, second or third feature of the system device of the present invention is arranged for each IP flow. 1 of the packets are set to be adjustable for the entire distribution of the hop count values.
Each of the packets in the plurality of internal nodes is provided with a functional unit that determines the required packet transfer priority stepwise so that the packet is distributed to any of two or more corresponding queues divided by the above thresholds. In the packet transfer control system, the transfer control means is provided with functional means for performing the required packet transfer scheduling using the two or more queues.

[0033]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will now be described with reference to the accompanying drawings in order of an apparatus example, a method example corresponding thereto, and a modified example (system apparatus example and a method example corresponding thereto). I will give you an explanation.

(Device Example) First, FIG. 1 is a system configuration diagram of an IP network to which a packet transfer control device according to the device example of the present invention is applied.

As shown in the figure, the IP network α to which the packet transfer control device 1 according to this device example is applied is
A plurality of nodes 2, 2, ...
The packet transfer control device 1 is installed in any one area on the main backbone line 4 in the IP network α and is used. It

FIG. 2 is a functional block diagram of the packet transfer control device 1 shown in FIG.

As shown in the figure, the present packet transfer control device 1 includes a packet receiving unit 5, a header analyzing unit (“header analyzing means” in the present invention) 6, and a hop count information learning / accumulating unit (main). According to the invention, "hop count information learning
(Storage means) 7), transfer priority determining unit (“transfer priority determining unit” in the present invention) 8, packet transfer control unit (“packet transfer control unit” in the present invention) 9, and packet transmitting unit 10 and.

Here, the packet receiving unit 5 receives a packet forming the IP flow generated in the backbone line 4 to generate a copy, and the copied packet is transferred to the header analyzing unit 6 and the transfer priority determining unit. It is given to 8 and.

Further, the header analysis unit 6 analyzes the header information of the packet given from the packet reception unit 5, and based on the header information, at least the source IP address and the destination IP address (hereinafter, both are collectively referred to as " Source / destination IP address), arbitrary mounting system dependent information, and the TTL value of the current packet (hereinafter referred to as "observed TTL value"), and all of these are acquired by the hop count information learning / accumulation unit 7 And the source / destination IP address to the transfer priority determining unit 8.

The mounted system dependent information referred to here is arbitrary information that can take a unique value depending on the type of OS (operating system) mounted on the host (terminal device 3) that issued the packet. As the specific mounting system dependent information, for example, the window size or the like can be applied.

On the other hand, the hop count information learning / accumulation unit 7
Estimates the type of the host OS installed from the mounting system dependency information acquired by the header analysis unit 6 and derives the initial TTL value initially assigned to the corresponding packet.
Based on the initial TTL value related to this derivation and the observed TTL value acquired by the header analysis unit 6, the hop count value of the packet between the two hosts that are the main constituents of the IP flow is estimated and calculated. The hop count value is accumulated for each corresponding IP flow with learning over time (“learning” means that estimation of the hop count value and registration based on it are repeated for each IP flow). This means that the contents of each database (total accumulated amount of hop count values) described later are enhanced).

Here, the hop count information learning / accumulating unit 7 implements the learning and accumulating function of the hop count value from three types of databases, that is, the mounting system dependent information acquired by the header analyzing unit 6. In order to derive the initial TTL value, the estimated installed OS type and initial TT
System-specific initial TTL database 7a in which the correspondence with the L value is set in advance, and this system-specific initial TTL
The TTL database 7b for registering the initial TTL value derived from the database 7a and the observed TTL value acquired by the header analysis unit 6 in association with each corresponding host IP address (see FIG. 3 (TTL shown in FIG. 2)). (See the figure showing an example of data registered in the database 7b)), and the hops approximately calculated using a predetermined calculation formula in which both the initial TTL value and the observed TTL value registered in the TTL database 7b are variables. Count value, corresponding to each IP
A flow database 7c (see FIG. 4 (see FIG. 4 showing an example of data registered in the flow database 7c shown in FIG. 2)) for registering in association with a source / destination IP address of a flow. .

The calculation formula used to calculate the hop count value is the terminal device (3, 3) of each of the hosts A and B.
If the initial TTL value corresponding to 3) is Ti _A , Ti _B and the observed TTL value acquired by the header analysis unit 6 is T _A , T _B , the packet transfer path between the hosts AB is symmetric (same transmission / reception path). ), The desired hop count value h
(fl _{AB) is, h (fl AB) = (} Ti A -T A) + (Ti B -
T _B ) +1.

Further, the hop count information learning / accumulation unit 7 takes into consideration the change in the hop count value due to the reason such as the route change characteristic of the IP network α, and at each arrival of the period set to be adjustable. , TTL database 7b and flow database 7c are also updated.

On the other hand, based on the hop count value accumulated in the flow database 7c in the hop count information learning / accumulation unit 7, the transfer priority determination unit 8 has a lower rank for an IP flow having a smaller hop count value. The packet transfer priority defined as follows is selectively determined by using the source / destination IP address provided from the header analysis unit 6 as a key, and more specifically, for each IP flow provided from the packet reception unit 5. Packets of the W are divided by one or more thresholds (parameters: h ₁ , h ₂ , ..., H _n ) that are set to be adjustable with respect to the entire distribution of hop count values. ₁ , W ₂ , ...
W _{n + 1} ) so that the required packet transfer priority is determined by, for example, 0 ≦ h (fl _AB ) <
It has a stepwise function means such as “lowest priority” for h ₁ , “low priority” for h ₁ ≦ h (fl _AB ) <h ₂ , and “highest priority” for h (fl _AB ) ≧ h _n. To do.

The packet transfer control unit 9 performs packet transfer scheduling according to the packet transfer priority determined by the transfer priority determination unit 8. More specifically, the required packet transfer scheduling is performed by the packet transfer Two or more queues (W ₁ ,
W ₂ , ... W _{n + 1} ) is provided.

(Example of Method) Next, a specific example of a packet transfer control method implemented by the example of the apparatus configured as described above will be described.

First, in this method example, as is clear from the configuration of the packet transfer control device 1 according to the above-described device example, header information analysis, hop count value calculation, accumulation and update, packet transfer priority The determination and the packet transfer scheduling are sequentially and collectively performed in an arbitrary area on the backbone line 4.

That is, first, the header analysis unit 6 in the packet transfer control device 1 analyzes the header information of the packets forming the IP flow generated in the backbone line 4, and then the hop count information learning / accumulation unit 7 On the basis of the header information analyzed by the header analysis unit 6, while presumably calculating the hop count value of the packet between the two hosts, which are the main constituents of the IP flow, and corresponding the calculated hop count value. The “hop count information learning process” for accumulating each IP flow is executed.

Here, FIG. 5 is a flow chart for explaining the hop count information learning processing executed in the hop count information learning / accumulation unit 7 shown in FIG.

As shown in the figure, when the hop count information learning process is performed, the hop count information learning / accumulation unit 7 first sends / receives the IP address given from the header analysis unit 6.
It is determined whether or not the hop count value corresponding to the address is registered in the flow database 7c (ST1.
1).

As a result of this determination, when the required hop count value is already registered in the flow database 7c (ST11; Yes), the hop count information learning / accumulation unit 7 immediately terminates the processing, but the hop count value concerned. Is not registered yet (ST11; N
o), it is determined whether or not the initial TTL value corresponding to the source / destination IP address is registered in the TTL database 7b (ST12).

Here, as a result of the above discrimination, when the initial TTL value relating only to the source IP address is registered (S
T12; right), the hop count information learning / accumulation unit 7
Although it is determined that the response packet from the destination IP address corresponding to the source IP address has not been obtained yet, the processing is immediately terminated, but the initial T
When the TL value is not registered (ST12; middle),
It is determined that the originating IP address has been acquired for the first time this time, and further it is determined from the system-specific initial TTL database 7a whether or not the initial TTL value corresponding to the originating IP address can be estimated (ST13).

As a result of this determination, if the required initial TTL value cannot be estimated even by referring to the system-specific initial TTL database 7a (ST13; No), the hop count information learning / accumulation unit 7 continues the processing. However, if the initial TTL value based on the system-specific initial TTL database 7a can be estimated (ST13; Yes), the initial TTL value and the observation corresponding to the originating IP address are observed. The TTL value is registered in the TTL database 7b (ST14).

Next, the hop count information learning / accumulation unit 7
Is the initial TTL value and observed TT for the originating IP address
After the L value is registered, it is determined whether or not the initial TTL value corresponding to the destination IP address is registered in the TTL database 7b (ST15).

Here, as a result of the above discrimination, when the initial TTL value for the destination IP address is not registered (ST
15; No), the hop count information learning / accumulation unit 7
When it is determined that the response packet from the destination IP address corresponding to the source IP address has not been obtained yet, the process is immediately terminated, but when the initial TTL value related to the destination IP address is registered ( ST15; Ye
s), the hop count value corresponding to the IP flow specified by the source / destination IP address is calculated using the above-mentioned calculation formula (h
(fl _AB ) = (Ti _A −T _A ) + (Ti _B −T _B ) +1), which is registered in the flow database 7c (ST16), and the required hop count information learning processing ends.

On the other hand, as a result of the above-mentioned determination processing of ST12, when the initial TTL value relating to the source / destination IP address is registered (ST12; left), the hop count information learning / accumulating unit 7 causes the above-mentioned ST16. In the process, the hop count value corresponding to the IP flow is immediately registered in the flow database 7c, and the required hop count information learning process ends.

Next, while the packet transfer control device 1 constantly carries out the hop count information learning process,
Every time the adjustable period is reached, the contents of the TTL database 7b and the flow database 7c are updated under the control of the hop count information learning / accumulation unit 7, and in parallel with this, the flow database 7c is updated. Based on the accumulated hop count value, the "packet transfer priority determination process" for selectively determining the packet transfer priority defined so that the IP flow having a smaller hop count value has a lower rank is transferred. This is performed under the control of the priority determination unit 8.

Here, FIG. 6 is a flow chart for explaining the packet transfer priority determining process executed in the transfer priority determining unit 8 shown in FIG.

As shown in the figure, when carrying out the packet transfer priority determining process, the transfer priority determining unit 8 can first specify the hop count value corresponding to the source / destination IP address from the flow database 7c. Determine whether or not (S
T21).

As a result of this determination, the flow database 7c
If the hop count value is registered in and the hop count value can be specified based on the flow database 7c (ST21; Yes), the transfer priority determining unit 8
Determines the designated value of the specified hop count value h (fl _AB ) (ST22).

Here, two parameters h ₁ are set as the thresholds that are set to be adjustable with respect to the entire distribution of the hop count values h (fl _AB ) registered in the flow database 7c.
And h ₂ is given in advance, along with this, three steps packet transfer priority defined in the transfer priority determining unit 8 so that its ranking smaller IP flows hop count value becomes low, h (fl _AB ) <H ₁ is “low”, h ₁ ≦ h (fl
It is assumed that when _AB ) <h ₂ is set to “standard”, and when h (fl _AB ) ≧ h ₂ is set to “high”.

At this time, the transfer priority determining section 8 responds to the instruction value of the hop count value h (fl _AB ) specified above.
When it is included in the range of h (fl _AB ) ≧ h ₂ (ST22; left), the required packet transfer priority is “high”.
(ST23), and when h (fl _{A B} ) <h ₁ is included in the range (ST22; medium), the packet transfer priority is set to “low” (ST24), and h ₁ ≦ h (f
When it is included in the range of l _AB ) <h ₂ (ST22;
Right), the packet transfer priority is set to "standard" (ST25), and the required packet transfer priority determination processing is ended.

On the other hand, as a result of the above-mentioned determination processing of ST21, if the required hop count value cannot be specified even by referring to the flow database 7c (ST21; N).
o), the transfer priority determining unit 8 sets the required packet transfer priority to "standard" in the above-described process of ST25, and ends the required hop count information learning process.

Then, the transfer priority determining unit 8 gives the information indicating the packet transfer priority determined as described above to the packet transfer control unit 9 together with the corresponding packet.

Next, the packet transfer control unit 9 executes the packet transfer scheduling according to the packet transfer priority given from the transfer priority determination unit 8. ,"standard",
Three queues W ₁ , W ₂ and W respectively corresponding to “high”
₃ is set in advance, and when the given packet transfer priority is “low”, the corresponding packet is queued W
₁ and if it was “standard”, the packet is stored in queue W ₂ , and if it was “high”, the packet is stored in queue W ₃ .

Then, the packet transfer control unit 9 assigns the packets distributed to the _three queues W ₁ , W ₂ and W ₃ as described above to the packet transfer scheduling according to the corresponding packet transfer priority. Packets given to the transmission unit 10 are sequentially sent out to the backbone line 4 in accordance with the scheduling, and as a result, the setting at the time of application is extremely simple, but fair to each IP flow on the IP network. As a result of effective bandwidth allocation, efficient packet transfer control can be realized as a whole.

(Modified Example) Next, a modified example of the system apparatus example of the present invention and the corresponding method example will be described.

FIG. 7 is a system configuration diagram of an IP network to which a packet transfer control system device according to a modification of the present invention is applied.

As shown in the figure, the IP network β to which the packet transfer control system device (not shown) according to the present system device example is composed of routers and the like, and includes the backbone line 4 (shown in FIG. 1). Equivalent) and a plurality of boundary nodes 2 at the connection points of a plurality of peripheral circuits
, a plurality of internal nodes 2 that integrate the plurality of boundary nodes 2a, 2a, ... On the backbone line 4.
b, 2b, ... And other plural peripheral nodes 2c, 2
, and terminal devices 3, 3, which respectively form a host.
... (equivalent to that shown in FIG. 1), the packet transfer control system device has the same or equivalent functional means as those in the above-mentioned device example, and has the plurality of boundary nodes 2a, 2a ,. Upper and multiple internal nodes 2b, 2
b, ... Dispersively arranged on the upper side.

That is, in each of the plurality of boundary nodes 2a, 2a, ..., The header analyzing unit (6), the hop count information learning / accumulating unit (7), and the transfer priority determining unit (8) in the above-described device example are respectively provided. ) Is equipped with all the functions of
Priority information tagging means (not shown) that tags (tags) the packet for each corresponding IP flow with priority information representing the packet transfer priority determined by the transfer priority determining unit (8). Are also provided.

The hop count information learning / accumulating unit (7) in the plurality of boundary nodes 2a, 2a, ... Has the system-specific initial TTL database (7a), TTL database (7b) and In addition to setting the flow database (7c), the hop count values registered in the flow database (7c) are shared by all of the plurality of boundary nodes 2a, 2a, ... Functional means for newly synchronizing the contents of the TTL database (7b) and the flow database (7c) in 2a, ...

On the other hand, a plurality of internal nodes 2b, 2
, b, ... Priority information tagged on the packets by the priority information tagging means in the plurality of boundary nodes 2a, 2a, ... As a modification means of the packet transfer control unit (9) in the above-mentioned device example. Based on
A packet transfer control means for performing packet transfer scheduling according to the corresponding packet transfer priority is provided.

According to the example of the system device configured as described above, the header information is analyzed, the hop count value is calculated,
Accumulation and update, determination of packet transfer priority and tagging are performed in a distributed manner at the plurality of boundary nodes 2a, 2a, ... And required packet transfer scheduling is performed at the plurality of boundary nodes 2a, 2a ,. , The internal nodes 2b, 2b, ... Are distributed in a distributed manner based on the priority information tagged in the packet. Therefore, the configuration itself is slightly complicated as compared with the case of the above-described device example. However, due to the effect of distributed processing, required packet transfer control can be performed at higher speed.

Further, according to the above system device example, the hop count value is shared by the plurality of boundary nodes 2a, 2a, ... As needed, so that the hop count is more efficient than in the case of the device example described above. The count value can be learned.

The example of the apparatus, the example of the method corresponding to the apparatus, and the modification of the embodiment of the present invention have been described above.
The present invention is not necessarily limited to the above-described means and methods, and can be appropriately modified and implemented within a range that achieves the object of the present invention and has the effects described below.

[0077]

As described above in detail, according to the present invention, fair bandwidth allocation can be performed for each IP flow on the IP network, although the setting at the time of application is extremely simple. As a result, efficient packet transfer control can be realized as a whole.

[Brief description of drawings]

FIG. 1 is a system configuration diagram of an IP network to which a packet transfer control device according to an example of the present invention is applied.

FIG. 2 is a functional configuration diagram of the packet transfer control device shown in FIG.

FIG. 3 is a diagram showing an example of data registered in a TTL database shown in FIG.

4 is a diagram showing an example of data registered in the flow database shown in FIG.

5 is a flowchart for explaining a hop count information learning process performed in the hop count information learning / accumulation unit shown in FIG.

FIG. 6 is a flowchart for explaining a packet transfer priority determination process executed by the transfer priority determination unit shown in FIG.

FIG. 7 is a system configuration diagram of an IP network to which a packet transfer control system device according to a modification of the present invention is applied.

[Explanation of symbols]

α, β ... IP network (applied system configuration) 1 ... Packet transfer control device 2 ... node 2a ... boundary node 2b ... internal node 2c ... peripheral nodes 3 ... Terminal device 4 ... Backbone line 5 ... Packet receiver 6 ... Header analysis unit 7 ... Hop count information learning / accumulation unit 7a ... Initial TTL database by system 7b ... TTL database 7c ... Flow database 8 ... Transfer priority determination unit 9 ... Packet transfer control unit 10 ... Packet transmission unit

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Shozo Naito             2-3-1, Otemachi, Chiyoda-ku, Tokyo             Inside Telegraph and Telephone Corporation F term (reference) 5K030 GA03 GA13 HA08 HB17 KA07                       KX29 LA03 LC01 MB07

Claims (17)

[Claims] 1. On a backbone line of an IP network, header information of a packet forming an IP flow generated in the backbone line is analyzed, and the IP flow generation entity is based on the analyzed header information. The hop count value of the packet between two hosts is presumedly calculated, and the calculated hop count value is set to the corresponding I
By accumulating for each P flow, and selectively determining a packet transfer priority defined such that the IP flow having a smaller hop count value has a lower rank based on the accumulated hop count value. A packet transfer control method comprising performing packet transfer scheduling according to the packet transfer priority. 2. The initial TTL of the packet, which is presumably derived from arbitrary mounting system-dependent information that can be obtained from the header information, in the calculation of the hop count value.
The packet transfer control method according to claim 1, wherein the packet transfer control method is performed based on a value and an observed TTL value of the packet that is directly acquired from the header information. 3. The calculation of the hop count value is approximately performed using a predetermined calculation formula in which both the initial TTL value and the observed TTL value are variables. Packet transfer control method. 4. The packet transfer control method according to claim 1, 2 or 3, wherein the hop count value is updated each time an adjustable period is set. 5. The packet transfer scheduling is performed using two or more queues divided by one or more thresholds set to be adjustable with respect to the entire distribution of the hop count value, and the packet transfer priority is set. Is determined by the packet corresponding to each of the IP flows.
The packet transfer control method according to claim 1, wherein the packet transfer control method is performed stepwise so that the packet is distributed to any of the above queues. 6. The header information analysis, the hop count value calculation, storage and update, the packet transfer priority determination, and the packet transfer scheduling are collectively performed in an arbitrary area on the backbone line. The packet transfer control method according to claim 1, 2, 3, 4, or 5. 7. The header information analysis, hop count value calculation, storage and update, and packet transfer priority determination are distributed at a plurality of boundary nodes at connection points between the backbone circuit and a plurality of peripheral circuits. The packet transfer scheduling is performed in a distributed manner in a plurality of internal nodes that integrate the plurality of boundary nodes on the backbone line. 5. The packet transfer control method described in 5. 8. The determination of the packet transfer priority in the plurality of border nodes is performed by a process of tagging the packet of each corresponding IP flow with priority information indicating the packet transfer priority. The packet transfer control according to claim 7, wherein the packet transfer scheduling in the plurality of internal nodes is performed based on the priority information tagged in the packet in the plurality of boundary nodes. Method. 9. The packet transfer control method according to claim 7, wherein the hop count value is shared by all of the plurality of boundary nodes at any time. 10. A packet transfer control device installed in an arbitrary area on a backbone line of an IP network, wherein header information of a packet forming an IP flow generated on the backbone line is analyzed and the header is analyzed. From the information,
Header analysis means for acquiring arbitrary mounting system dependent information and observed TTL value, and from the mounting system dependent information acquired by this header analyzing means, presumably derive an initial TTL value of the corresponding packet, and The initial TT related to this derivation
L value and the observation T acquired by the header analysis means
Based on the TL value, it is the originator of the IP flow 2
A hop count information learning that predictively calculates the hop count value of the packet between two hosts, and accumulates the hop count value related to this calculation with learning over time for each corresponding IP flow. Based on the storage means and the hop count value accumulated in the hop count information learning / accumulation means, the packet transfer priority defined so that the IP flow having a smaller hop count value has a lower rank is selectively selected. A packet transfer control device, comprising: transfer priority determining means for determining; and packet transfer control means for performing packet transfer scheduling according to the packet transfer priority determined by the transfer priority determining means. . 11. The hop count information learning / accumulating means includes a system-specific initial TTL database preset for deriving the initial TTL value from the mounting system-dependent information acquired by the header analyzing means, and this system. A TTL database for registering the initial TTL value derived from another initial TTL database and the observed TTL value acquired by the header analyzing means in association with each corresponding host IP address, and registering in this TTL database And a flow database for registering the hop count value, which is approximately calculated using a predetermined calculation formula in which both the initial TTL value and the observed TTL value that have been set are variables, in association with each corresponding IP flow. The package according to claim 10, further comprising: Transfer controller. 12. The hop count information learning / accumulation unit is configured to perform the TT count every time an adjustable period is reached.
The packet transfer control device according to claim 11, further comprising a functional unit that updates the contents of the L database and the flow database. 13. The transfer priority determining means, wherein the packets for each of the IP flows are classified by one or more thresholds set to be adjustable with respect to the entire distribution of the hop count values. The packet transfer control means is provided with functional means for stepwise determining the required packet transfer priority so that the packet transfer scheduling can be distributed to any of the above queues. 13. The packet transfer control device according to claim 10, 11 or 12, further comprising: a functional unit that executes using the. 14. A packet distributedly arranged at a plurality of boundary nodes at a connection point between a backbone circuit of an IP network and a plurality of peripheral circuits, and a plurality of internal nodes that integrate the plurality of boundary nodes on the backbone circuit. A transfer control system device, wherein the plurality of boundary nodes analyze header information of a packet forming an IP flow generated in the backbone line, and from the header information,
A header analysis unit that acquires arbitrary mounting system dependent information and an observed TTL value, and presumably derives an initial TTL value of the corresponding packet from the mounting system dependent information acquired by the header analyzing unit, The initial TT related to this derivation
L value and the observation T acquired by the header analysis means
Based on the TL value, it is the originator of the IP flow 2
A hop count information learning that predictively calculates the hop count value of the packet between two hosts, and accumulates the hop count value related to this calculation with learning over time for each corresponding IP flow. Based on the storage means and the hop count value accumulated in the hop count information learning / accumulation means, the packet transfer priority defined such that the IP flow having a smaller hop count value has a lower rank is selectively selected. Transfer priority determining means for determining, and priority information tagging means for tagging the priority information indicating the packet transfer priority determined by the transfer priority determining means to the packet of each corresponding IP flow. , And each of the plurality of internal nodes has the priority information tagging procedure at the plurality of boundary nodes. A packet transfer control system device, comprising: packet transfer control means for performing packet transfer scheduling according to the corresponding packet transfer priority based on the priority information tagged by the stage. 15. The hop count information learning / accumulating means in each of the plurality of boundary nodes is preset by system for deriving the initial TTL value from the mounted system dependent information acquired by the header analyzing means. A TTL database, and a TTL database for registering the initial TTL value derived from the system-specific initial TTL database and the observed TTL value acquired by the header analysis means in association with each corresponding host IP address , The hop count value that is approximately calculated using a predetermined calculation formula in which both the initial TTL value and the observed TTL value registered in this TTL database are variables is registered in association with each corresponding IP flow. And a flow database for The hop count information learning / accumulating means stores the contents of the TTL database and the flow database in the plurality of boundary nodes so that the hop count value registered in the flow database is shared by all of the plurality of boundary nodes at any time. 15. The packet transfer control system device according to claim 14, further comprising functional means for synchronizing each other. 16. The hop count information learning / accumulating means in each of the plurality of boundary nodes receives the TT every time when a period set to be adjustable is reached.
The packet transfer control system apparatus according to claim 15, further comprising functional means for updating the contents of the L database and the flow database. 17. The transfer priority determining means in the plurality of boundary nodes is configured such that the packet for each of the IP flows is set to one or more thresholds that are set to be adjustable with respect to the entire distribution of the hop count value. The packet transfer control means in each of the plurality of internal nodes comprises functional means for stepwise determining the required packet transfer priority so that the packet transfer priority can be distributed to any of two or more corresponding queues. The packet transfer control system device according to claim 14, 15 or 16, further comprising functional means for performing the required packet transfer scheduling using the two or more queues.