JP2001007854A - System and method for reducing average delay time in packet transfer network - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce an average delay time for frame transfer by storing an optimum frame length in advance in response to the attribute of a user packet from each subscriber network and building up a frame with a selected specified frame length corresponding to the attribute that is provided to a user packet and read. SOLUTION: An edge node 100 receives a user packet from a subscriber network and transfers the packet to a post-stage as a super-frame and is provided especially with an optical wave ADP header generating section 14, a storage section 40 that stores information of a specified frame length acquired from the received user packet, and an IP packet priority read section 12 reading the information of the received packet required to decide the frame length. An observation section 20 of a core node 200 observes the number of transfer packets per unit time in the unit of ports in the core node 200 or the number of transfer packets per unit time in the unit of nodes in the core node.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a lightwave network data communication system in which a user packet from a subscriber is accommodated in a frame having a predetermined length or more and is transferred within a lightwave network, and a transfer delay in a core node passing through the lightwave network is reduced. The present invention relates to an average delay time reducing method and method in a packet transfer network.

[0002]

2. Description of the Related Art In recent years, realization of a high-speed and large-capacity backbone network corresponding to multimedia data such as the Internet has been demanded.
Promising is a data transfer technology for lightwave networks using M (wavelength division multiplexing) technology.

FIG. 6 is an explanatory diagram showing a configuration of a conventional lightwave network. The lightwave network includes an edge network accommodating a subscriber network, as shown in FIG.
It consists of a core network that provides the main transfer function of the lightwave network. Although FIG. 1 shows a case where the edge network and the core network have a mesh configuration, a network configuration other than the mesh configuration may be employed.

The edge node constituting the edge network reads the destination information of the user packet received from the subscriber network, converts the protocol of the user packet, and converts it into a frame for lightwave network transfer as a core node constituting the core network. Send to.

The core node routes the received frame to an appropriate destination, and outputs the frame from an ultra-high-speed port for transferring data to a subsequent node.

However, even when the WDM technique for providing a high-speed and large-capacity transfer function is used, an arbiter time (arbitration time) which is a switch interval in a core node is used.
Is limited by the number of output ports, etc.
It is difficult to shorten.

That is, when a user packet having a short packet length is directly transferred into an optical network, the processing speed of a packet switch included in a core node becomes a bottleneck even if a data transfer function by WDM using a high-speed wavelength path is used. However, there is a problem that desired performance is not obtained.

More specifically, a general packet switch used in a core node is an input buffer type crossbar switch. As a general problem of the input buffer type switch, data to be output in a timing conflicts. There is HOL (Headof Line) blocking. For this reason, in the present input buffer type switch, the processing throughput of the entire core node device reaches a plateau even if the switch switching performance is improved.

For example, even if a switch section of 80 Gbps is constructed, the processing throughput will only increase to substantially 58% unless HOL is eliminated, and 80 Gbps × 0.58
= 46 Gbps.

As a method of solving the problem of the HOL, a method of preparing a virtual packet queue for each output port in an input buffer, and judging another transferable output port when a HOL occurs, and transferring the data to the output port is determined. is there. However, even with such a processing method, there is a limitation due to the scheduling of the transfer data, and with the current electric processing technology, Equation 1
The switch capacity of 0 Gbps becomes the limit.

As described above, there is a limit in shortening the switch interval of the packet switch of the core node.
In order to efficiently transfer data and improve processing throughput in a core network composed of core nodes having such a packet switch unit, it is necessary to make the transfer frame longer based on the switch interval. It is valid.

Under such circumstances, as a technique for increasing the transfer efficiency by constructing a plurality of user packets into a frame for transmission over a lightwave network having a certain length or more, the applicant has filed Japanese Patent Application No. Hei.
A patent application for the lightwave network of 1-054400 has been filed.

In this prior art, as shown in FIG. 6B, a protocol stack configuration of a lightwave network according to the prior art is shown.
Is a synchronous digital network (SDH) or synchronous optical network (SONET) layer, and has a protocol layer that defines a lightwave adaptation layer (lightwave ADP layer) as an intermediate layer of a network layer that performs IP packet transfer.

The main functions of the lightwave adaptation layer are:
It has the function of address resolution of the lightwave header, switching of lightwave adaptation frames, and construction of a lightwave adaptation frame for transferring a lightwave network to a lightwave network, and conversely, a function of disassembling the lightwave adaptation frame into a user packet.

At the edge node receiving the user packet from the subscriber network, a long packet having a packet length longer than a certain length is transferred to a subsequent stage as it is as a single lightwave adaptation frame, and the packet length is short. As for packets, a plurality of packets are assembled into one lightwave adaptation frame and transferred within the lightwave network. The light adaptation frame having a certain length or more is defined as a superframe.

The relationship between the arbiter time and the frame length of the prior art using such a superframe will be described with reference to FIG.

FIG. 9 is an explanatory diagram showing the relationship between the arbiter time and the frame length. As shown in FIG. 9, when a normal packet received from the non-superframed subscribed network arrives at the packet switch unit in the node, the packet processing unit in the node reads one packet at a time scheduled at the arbiter time. To the destination port, in this case, in units of one arbiter time,
Since one packet transfer is performed, if the amount of received packets is large, a packet output wait occurs in the packet switch unit, and the data transfer rate deteriorates.

However, when a superframed frame arrives, the transfer interval scheduled by the arbiter time can be effectively used, and the transfer efficiency is improved.

That is, when a plurality of packets having the same destination and the same type are accommodated in a frame of a certain length or more and transferred within the lightwave network as a superframe, the transfer of a plurality of user packets can be substantially performed in one arbiter time. You can do it.

In the prior art, in order to keep the utilization rate of the payload in the superframe at a certain level or more, the passing nodes combine the packets. This method is called a share ride (share ride). By superimposing a packet having the same destination with respect to a super packet having a low payload utilization rate, the packet is transferred as a super frame for accommodating a packet of a certain length or more. The aim is to improve the transfer efficiency in a lightwave network.

For example, focusing on a certain traffic flow, a data transfer method in a lightwave network having a share / ride function in the prior art will be specifically described with reference to FIGS. 6 and 7. FIG.

I entered from edge node c1 shown in FIG.
The P packet is an edge node c4 accommodating the subscriber network D
If the edge node c1 has
From the destination address and service class of the received packet, the address of the lightwave header is resolved, aggregated into a fixed unit and converted into a lightwave adaptation frame (converted to a single frame, then converted to a superframe) and separately routed. Send to the determined transfer destination. Here, c1 → b1 → a1 → d1 → core node 1
→ The core node 5 → core node 4 → a4 → b4 → c4, and the route is transferred. In the edge nodes b1 and a1 which pass through, the subscriber network B and the subscriber are respectively at the same timing as the packet. A case where a user packet having the same service class as the same destination c4 is transmitted from the subscriber network from the network C is illustrated.

Here, a configuration is shown in which all edge nodes are terminated at the lightwave ADP layer.
In some embodiments, termination is performed at the layer and superframe construction / disassembly and superframe-based switching are not performed.

Next, the construction operation of the super frame will be described with reference to FIG. FIG. 7 is a block diagram showing a superframe transfer operation according to the related art. As shown in FIG. 7, the IP packet received by the originating edge node c1 is converted from the read destination address and the service class into a single frame which is a frame to which an appropriate lightwave header is added (Process 1). , And c
1, a plurality of single frames belonging to the same aggregated flow are constructed as a superframe having a long frame length, and transferred to the subsequent node b1 (process 2).

Next, the edge node b1 receiving the superframe waits for the arrival of a subsequent packet belonging to an aggregated flow having the same destination and the same type by a predetermined timeout value, and converts the received packet into a single frame. At the same time, the obtained single frames are accommodated in the payload of the superframe (process 3).

When a timeout period for waiting for reconstructing has elapsed from the reception of the superframe, or when the superframe has reached a prescribed frame length, the new packet is converted into a single frame to the subsequent edge node, and Join the frame payload and transfer the reconstructed superframe.

Then, the next edge node a1 receiving the superframe converts the IP packet having the same destination and the same type received at a1 if the packet accommodation ratio of the arriving frame does not reach the specified value. Further, the superframe construction is restarted (process 4).

In the manner described above, this super frame is
While increasing the capacity of packets to be stored in the payload,
The data is transferred to the edge node c4 which is the final destination.
In this way, data transfer by a super frame having a high packet accommodation ratio can be realized.

However, if this superframe length is simply set to be long, the packet load can be reduced and the transfer efficiency can be increased, but there is a problem that the superframe construction time in the edge network becomes longer and the average delay time increases. .

Therefore, in the prior art, the share
In order to guarantee the maximum delay time during data transfer using the ride method, the maximum number of nodes that can wait for a share ride and the timeout value are
The condition is set as a condition for canceling the ride process, and in the transit node after satisfying the condition, the transfer process is performed as it is without waiting for the share ride.

FIG. 8 shows an embodiment in which a superframe construction wait condition is added to the share ride in the prior art. The originating edge node sets the number of routers for superframe construction waiting and the construction wait timeout value for each received user packet at the transit core node or the edge node. Does not perform the share / ride processing, which is the waiting of the same type of frame, for reconstructing the frame, and the subsequent nodes simply perform the switching processing and transfer the data directly to the subsequent stage.

Here, the number m of nodes for which a share ride wait is performed in the header portion of the lightwave adaptation frame.
Set. Then, among the transit nodes located on the transfer path, as a result of the share ride wait processing, the node which has failed to construct a superframe longer than the prescribed frame length within the prescribed wait time and has timed out, decrements the number of share ride waits. After that, the packet is transferred to the next-stage core node. Then, in the nodes following the transit node in which the number of shared ride waits has detected "0", the shared ride wait is not performed even if the superframe is less than the specified length.
Perform pass-through transfer.

Also, the applicant has filed Japanese Patent Application No. 11-001441.
Of the centralized communication network observation control device for which
The prior art will be described with reference to FIG. Japanese Patent Application No. 1
As disclosed in Japanese Patent Application Laid-Open No. 1-001441, there is a method of monitoring the upper layer and controlling the link capacity and link path of the lower layer.

In this prior art, an observation unit 91 which takes in the state from a node constituting the communication network 90 to be observed / controlled periodically on a link basis or as an observation value when a certain period of time has passed. It has a storage unit 92 for storing information on each link including the observation value within the designated period and the service content, and the calculation unit 94 determines the lower order from the storage content of the storage unit 92 and the information of the control content 93 of each layer. The configuration information of the lower layer including the link capacity and the link route is determined in consideration of whether or not the service content can be satisfied when the layer is controlled.

[0035]

SUMMARY OF THE INVENTION However, Japanese Patent Application No.
The method disclosed in Japanese Patent Application No. 054400 has the following improvements.

The first point is that only the maximum delay time for each aggregated flow can be guaranteed only by the maximum number of nodes that can wait for a share / ride and the timeout value that is the maximum wait time for superframe reconstruction at one node.

That is, the delay time substantially varies greatly depending on the traffic volume for each aggregated flow, and when the traffic volume in the network decreases, the delay for constructing the superframe increases. Although the packet load of the core network is small, an attempt is made to construct a long frame, so that the average delay time increases and approaches the maximum delay time, which is not efficient.

Further, when the waiting time increases at each node, the required memory amount in the lightwave router increases. For this reason,
At the time of setting the aggregation flow, the prescribed frame length should be statically selected according to the expected traffic volume and the expected traffic load condition of the core network, but there is no such method in the conventional method.

The second point is that the specified frame length does not correspond to a dynamic change in traffic. It is difficult to predict traffic characteristics in multimedia communications having various speeds such as recent voice communication, image communication, and data communication. Therefore, it is useful to observe at regular intervals. Regularly changing the specified frame length based on the service contract information, the observed traffic volume per aggregated flow, and the traffic load on the core network can be expected to reduce the average delay time. Absent.

Further, in the system based on the centralized communication network observation control device of Japanese Patent Application No. 11-001441, there is no description about control of a specified frame length which characterizes a lightwave network.

[0041]

SUMMARY OF THE INVENTION The present invention utilizes wavelength division multiplexing (WDM) technology in the lower layer of synchronous digital hierarchy (SDH) to convert user packets from subscribers and forward them within a lightwave network. A lightwave adaptation layer to be built in a lightwave adaptation frame for performing a transfer passing per unit time in a lightwave network including an edge node accommodating a subscriber network and a core node performing a relay transfer process in the core network. It monitors the number of packets and monitors the entire network with a core node having a means for outputting the observation information,
From the observation information received from the core node, a central monitoring device having means for determining the traffic load status of the network and outputting a traffic load status determination result, based on the traffic load status determination result received from the central monitoring device An edge node having means for determining a frame length of a light adaptation frame accommodating a user packet from a subscriber network, wherein the edge node further comprises, according to an attribute of a user packet from each subscriber network, A storage unit for storing an optimum frame length in advance, attribute identification means for reading an attribute given to a user packet received from the subscriber network, and a frame for selecting a frame length in accordance with the read attribute of the packet Length selection means and the selected specified frame It has, and a frame constructing means for constructing a lightwave adaptation frame and containing one or more of the user packets.

Further, the method and method for reducing the average delay time in the packet point mangling according to the present invention are not limited to the lightwave network as the applicable network, but are long in the upstream network in order to reduce the packet processing load on the downstream network. The present invention is applied to any packet transfer network in terms of aggregating packets into frames.

In other words, the present invention provides a subscriber access network including a subscriber access device for accommodating a subscriber network, receiving a user packet from the subscriber, and transferring the user packet to a subsequent data communication network; In a packet transfer network including a data transfer network connected to an access network and including a data transfer device that performs a relay transfer process of a packet, the packet transfer network converts a received user packet and stores the converted user packet in the packet transfer network. An adaptation layer to be constructed as a new frame for forwarding is defined, and the subscriber access device selects a frame length of a forwarding frame to be constructed in the packet forwarding network according to an attribute of a user packet received from the subscriber network. And transferring the data to the data transfer network.

The subscriber access device extracts a plurality of user packets belonging to the same attribute from the attribute information of the input user packet, and extracts attribute information relating to the extracted user packets and information of the data communication network. From the load situation, a frame length for transferring in the packet communication network including the data communication network is selected, and a new frame having the selected frame length, and the extracted user packet is constructed as payload information. Means.

[0045]

BRIEF DESCRIPTION OF THE DRAWINGS FIG.
No. 1-054400, a superframe constructed by an edge node in a lightwave network that performs a shared ride on a superframe with a plurality of packets connected in the order of arrival, accommodated in a payload, and attached with a common header. Is characterized in that a plurality of prescribed frame lengths can be arbitrarily set, and a function of selecting an optimal prescribed frame length according to the contracted network service quality and the expected traffic amount, and reducing the average delay time.

According to the present invention, for a user packet received from a subscriber network, for example, selection of a static prescribed frame length for selecting a different prescribed frame length for each service in advance, and at the time of high load and low load according to network conditions. This allows dynamic selection of a specified frame length for selecting a specified frame length.

As a result, if the superframe length is increased in the conventional technique, the packet load can be reduced and the transfer efficiency can be increased, but the superframe construction time in the edge network becomes longer and the average delay time increases. It is possible to solve the problem and reduce the average delay time of superframe transfer using the share ride of the lightwave network.

According to the present invention, the traffic amount and the packet load status of the aggregated flow are observed, and from the observation information, the possible specified frame length stored in the storage unit and the corresponding service contract contents are referred to, and It has a function of selecting a specified frame length and setting and controlling the super frame length to the selected specified frame length.

The mechanism for changing the specified frame length in the method and apparatus for reducing the average delay time in the lightwave network according to the present invention will be described below.

FIG. 1 is a block diagram showing a functional outline of an embodiment of the present invention. FIG. 2 is a block diagram showing a system configuration according to the embodiment of the present invention. As shown in FIG. 2, the system configuration of the present invention includes an edge node 100 forming an edge network and a core node 20 forming a core network.
And a central monitoring device 300 for monitoring a lightwave network formed by the core network and the edge network.

Next, an outline of the operation of changing the specified frame length in the embodiment of the present invention will be described with reference to FIG.

The core node 200 to be observed has an observation unit 20 for observing the load on the internal packet switch 21. Then, the central monitoring apparatus 300 determines a load condition based on the traffic observation data collected from the observation target node, and controls the related edge node 100 based on the determination unit to control the specified frame length. Has a control unit 32 for transmitting the control message.

The edge node 100 receives the user packet from the subscriber network and transfers it to the subsequent stage as a superframe. In particular, the lightwave ADP header generator 14 and the specified frame length information that can be taken for the received user packet are transmitted. A storage unit 40 for storing the information and an IP packet priority reading unit 12 for reading information of a received packet necessary for determining a frame length are mounted.

Next, the operation flow of the present invention will be described with reference to FIG. First, in the observation unit 20 of the core node 200,
The number of transfer packets per unit time at the core node per unit time or the number of transfer packets per unit time at the core node per node is observed.

Here, as an example of a specific method of calculating the load status of each node, (total number of packets transferred per unit time of all observation target nodes) / (maximum transfer per unit time of all observation target nodes) There is a calculation method based on the value of (total number of packets).

For the observation of transfer packets, it is optimal to observe the number of transfer packets per unit time and the average packet length for all ports of all nodes, but the cost and processing load of the observation device increase. In this embodiment, consider a case where some core nodes are selected as the traffic observation targets of the observation unit, and the number of transfer packets per unit time per node is observed.

Next, the determination unit 31 in the central processing unit 300 obtains, from the observation unit 20 of the core node, the number of collected transfer packets per unit time per node.
Then, the load status of each node is determined, and, for example, the state of whether the current core network is “high load” or “low load” is notified to some or all edge nodes in the network via the control unit 32. . Then, the edge node 100 that has received the notification from the control unit 32 stores the notified current load information in the storage unit 40.

The lightwave adaptation frame generator 14 in the edge node 100 determines whether the contents of the subscriber contract
If a service corresponding to the priority of the IP packet is desired, the priority read by the IP packet reading unit 12, the association information in the storage unit 40, the lightwave header information, and further input from the determination unit 31. A superframe is constructed based on information on the load state (high load or low load, etc.).

Next, FIG. 2 shows the overall configuration of the system of the present invention.
Next, a description will be given of the function allocation in the edge node with reference to FIG. 3 and the function allocation in the core node with reference to FIG.

FIG. 5 shows the arrangement of functions in a core node according to the present invention. The core node 200 in FIG. 5 includes a lightwave interface unit (not shown) corresponding to each wavelength path.
A packet switching unit 21 and an observation unit 20 for observing the load status of the packet switching unit 21.

The core node 200 transfers the received superframe via the packet switch 21 based on the information in the lightwave header. Here, the number of transfer frames per unit time in the packet switch unit 21 and the specified length are observed by the observation unit 20, and are notified to the determination unit 31 of the central monitoring apparatus 300.

Originally, the load of the core node to be observed should be observed for each port of the packet switch 21 or the like. In this example, an embodiment in which the observation is performed for each node is shown.

In the edge node shown in FIG. 3, the priority reading unit 12 reads the priority of the IP packet coming from the IP network as the subscriber network from the TOS value in the IPv4 header. In the present embodiment, the method of determining the priority of an IP packet includes a TOS value in an IP header, a port number in a TCP header, a physical port number in which an IP packet enters, and a physical port number in which an IP packet enters. A method such as determination based on a logical port number, for example, VCI and VPI values in ATM can be given. Here, as an example,
It is possible to use the TOS value in the IP header.

Next, based on the priority read from the IP packet, the packet load status data (for example, high load or low load) of the core network input from the determination unit 31 and the information in the storage unit 40, the specified frame is defined. The length selection unit 10 selects a specified frame length. Then, a frame having the selected specified frame length is generated, the user packet information is converted and included in the lightwave header, and the lightwave header adding unit 14 encapsulates the user packet.

As shown in an embodiment in FIG. 4, the storage unit 40 stores IP packet information, which is a received user packet, a service association file table 41 for the user packet, and a service association file for the contract port. , A service class defined for the lightwave adaptation frame as the lightwave header information table 43, a frame length at each time of high load / low load, a timeout value, a maximum number of nodes waiting for construction, and the like. .

With reference to the stored information, the specified frame length selection unit 10 can determine the specified length of the superframe that accommodates the received packet. For example, if the read priority of the received user packet is the value a, the service class of the lightwave ADP layer corresponding to the priority is searched with reference to the table 41, and the corresponding value A is obtained.

Next, by referring to the lightwave header information table corresponding to the service class A, the frame length corresponding to the load information notified from the determination unit is determined based on the specified frame length under high load or the specified frame under low load. By reading the specified frame length value, the timeout value, the maximum number of nodes waiting for superframe construction (SF construction wait), and the like described in the prior art, and the like, the user reads the frame length value. The packet is transmitted to the lightwave header providing unit 14. The packet at this point has a single-frame format to which a lightwave header is added but not superframed.

The single frame that has passed through the packet switch 11 is distributed by the dispatcher 15b to a superframe construction unit (SF construction unit) 17 corresponding to a predetermined transfer destination based on information in the lightwave header.

The single frame transferred to the superframe construction unit 17 waits for arrival of packets of the same priority at the same destination, which is an aggregation unit, until the specified frame length is reached, and connects them to construct a superframe. The scheduler 18 manages the superframe construction time based on the information in the lightwave header.

The superframe coming from the lightwave network is sorted by the dispatcher 15 based on the lightwave address. If the read lightwave address is not addressed to the own node, the lightwave address is transferred to the packet switch 11 as it is.

If the packet is a lightwave packet addressed to its own node, the superframe is disassembled by the superframe disassembly section 16, restored to a short original user packet, and then an optical wave header is added again and transferred to the packet switch 11. Then, after passing through the packet switch 11, the lightwave header removing unit 19 removes the header and transfers the header to the IP network.

Next, the method and method for reducing the average delay time in the packet transfer network according to the present invention are not limited to the lightwave network as the applicable network, but are used in the upstream network to reduce the packet processing load on the downstream network. It can be applied to any packet transfer network in terms of aggregating packets into long frames.

That is, the present invention can be applied to all cases where a communication network is configured using nodes such as routers having a low packet forwarding capability such as a small number of packets transferred per unit time [packets / sec].

For example, in the case of soft routing in which a node performs a routing process by software, the forwarding capability is reduced at the order level in the current technology as compared with the routing process by hardware, but the transmission capacity of the link is reduced. Is applicable when there is enough.

A subscriber access network including a subscriber access device for accommodating the subscriber network as a former stage network, connecting to the subscriber access network,
The present invention can be applied to a packet communication network including a data transfer network including a data transfer device that performs relay transfer of a received packet.

In this case as well, received packets are aggregated in the former stage network according to a certain aggregation process, for example, a system in which packets having the same destination and the same priority are aggregated, and a long frame is generated.

When the frame is lengthened in the pre-stage network and transferred to the post-stage network, the number of frames transferred in the post-stage network is reduced, and the load on the forwarding engine for executing the data transfer process is reduced. However, if the length is set too long, the waiting time for frame construction will increase, and the end-to-end transfer delay time will increase. If there is room in the load situation,
It is effective to control the length of the frame formed by the former net.

[0078]

The first point is that the average delay time can be reduced by changing the frame length from a single specified frame length to a plurality of frames. Only controlling the timeout period of the superframe and the maximum number of nodes waiting to construct the superframe could control only the maximum waiting time.

The second point is that the average delay time has fluctuated greatly in conjunction with traffic fluctuation, but the average delay time can be shortened by controlling the specified frame length to be short.

Further, when the traffic volume is small, not only is the specified frame length set short, but also the packet load and the situation of the core network can be controlled. (When the traffic volume is large, the specified frame length is set long.) This can be a mechanism for effectively using network resources and improving service quality.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a functional outline in an embodiment of the present invention.

FIG. 2 is a block diagram illustrating a system configuration according to an embodiment of the present invention.

FIG. 3 is a functional block diagram of an edge node according to the embodiment of the present invention.

FIG. 4 is an embodiment of a storage unit of an edge node according to the present invention.

FIG. 5 is a functional block diagram of a core node according to the embodiment of the present invention.

FIG. 6 is an explanatory diagram showing a lightwave network configuration according to a conventional technique.

FIG. 7 is a block diagram illustrating a superframe transfer operation according to the related art.

FIG. 8 is a block diagram showing a superframe construction waiting condition in the related art.

FIG. 9 is an explanatory diagram showing a relationship between an arbiter time and a frame length.

FIG. 10 is a functional block diagram of a second conventional technique.

[Explanation of symbols]

 Reference Signs List 10 Reference frame length selection unit 11, 21 Packet switch 12 IP packet priority reading unit 14 Lightwave header addition unit 15 Dispatcher 16 SF disassembly unit 17 SF construction unit 18 Scheduler unit 19 Lightwave header removal unit 20 Observation unit 31 Judgment unit 32 Control unit Reference Signs List 40 storage unit 41 IP packet-service association information 42 contracted port-service association information 43 lightwave header information 90 communication network to be observed / controlled 91 observation unit 92 storage unit 93 control contents of each layer 94 calculation unit 95 control unit 100 edge node 200 core node 300 central monitoring device

Claims (18)

[Claims] 1. A lightwave adaptation frame for converting a user packet from a subscriber into a lightwave adaptation frame by using wavelength division multiplexing (WDM) technology in a lower layer of a synchronous digital hierarchy (SDH). In the lightwave network including a lightwave adaptation layer to be built and including an edge node accommodating a subscriber network and a core node performing a relay transfer process in the core network, the edge node is configured by an attribute of a user packet received from the subscriber network. An average delay time in a packet transfer network, comprising: selecting a frame length, constructing the lightwave adaptation frame, and transferring the frame to the core node. 2. The edge node according to an attribute of a user packet from each subscriber network, a storage unit for storing an optimum frame length in advance, and an edge node attached to a user packet received from the subscriber network. Attribute identification means for reading an attribute, frame length selection means for selecting a frame length corresponding to the attribute of the read packet, and having the selected specified frame length, containing one or more of the user packets 2. The method as claimed in claim 1, further comprising frame construction means for constructing an optical adaptation frame. 3. The storage unit of the edge node includes a first information table that is determined in advance for a user and determines a service class based on attribute information given to a user packet, and is set for each service class. 3. The method according to claim 2, further comprising a second information table for storing a frame length corresponding to a traffic load condition in the lightwave network. 4. The method according to claim 2, wherein said frame length selecting means selects a frame length for each priority indicated by an attribute of the received user packet. 5. The method according to claim 2, wherein said frame length selecting means selects a frame length in accordance with a load situation in said lightwave network. 6. Utilizing wavelength division multiplexing (WDM) technology in a lower layer of synchronous digital hierarchy (SDH) to convert a user packet from a subscriber and construct a lightwave adaptation frame for transfer in a lightwave network. In the lightwave network including the lightwave adaptation layer and the edge node accommodating the subscriber network and the core node performing the relay transfer processing in the core network, the number of transfer packets passing per unit time is observed, and the observation information is obtained. A central node that receives the observation information output by the core node, determines the traffic load status, and outputs a traffic load status determination result, based on the determination result received from the central monitoring device. Optical adapter for accommodating user packets from the subscriber network An edge node for determining a frame length of an application frame. 7. A subscriber access network accommodating a subscriber network, including a subscriber access device for receiving a user packet from the subscriber and transferring the user packet to a subsequent data communication network, and connecting to the subscriber access network. And a packet transfer network including a data transfer network including a data transfer device that performs a packet relay transfer process. The packet transfer network converts a received user packet and transfers the converted user packet to the packet transfer network. An adaptation layer constructed as a new frame, wherein the subscriber access device selects a frame length of the frame for transfer of the packet transfer network according to an attribute of a user packet received from the subscriber network; An average delay time shortening method in a packet transfer network, comprising a configuration for transferring data to a data transfer network. 8. A subscriber access network accommodating a subscriber network, receiving a user packet from the subscriber, and transferring the packet to a subsequent data communication network. In a packet transfer network configured from a data communication network including a data transfer device that performs a relay transfer process of a received packet, the subscriber access device is configured to respond to a user packet attribute from the subscriber network,
A storage unit for storing an optimal frame length in advance; attribute identification means for reading attribute information given to user packets input from the subscriber network and extracting a plurality of user packets belonging to the same attribute; Attribute information about the extracted user packet; and
Frame length selecting means for selecting a frame length for transfer in the data communication network from the load state of the data communication network, and the extracted user packet having the selected frame length, and And frame constructing means for constructing a frame for subsequent data communication network transfer.
2. The average delay time reduction method in the packet transfer network according to 1. 9. A subscriber access network including a subscriber network, a subscriber access device connected to the subscriber network, receiving a user packet from the subscriber, and transferring the packet to a subsequent data communication network. In a packet transfer network consisting of a data transfer network including a data transfer device that connects to a user access network and transfers received packets, it observes the number of packets transmitted per unit time and outputs the observation information. A data transfer device, a central monitoring device that receives the observation information output by the data transfer device, determines a traffic load status, and outputs a traffic load status determination result, to the determination result received from the central monitoring device. A subscriber access device for determining a frame length of an optical adaptation frame accommodating a user packet from the subscriber network based on the subscriber network; And an average delay time reducing method in a packet transfer network. 10. An edge node accommodating a subscriber network,
In a lightwave network including a core node that performs a relay transfer process in a core network, the lightwave network uses a wavelength division multiplexing (WDM) technique in a lower layer of a synchronous digital hierarchy (SDH) to transmit user packets from a subscriber. And convert
The lightwave adaptation frame includes a lightwave adaptation layer to be built into a lightwave adaptation frame for transfer in a lightwave network, wherein the edge node selects a frame length according to an attribute of a user packet received from a subscriber network to build the lightwave adaptation frame. And transmitting the lightwave adaptation frame to the core node. 11. The edge node stores an optimum frame length in a storage unit in advance according to an attribute of a user packet from each subscriber network, and assigns the optimum frame length to the user packet received from the subscriber network. Read an attribute, select a frame length corresponding to the attribute of the read packet, and construct a lightwave adaptation frame having the selected specified frame length and accommodating one or more of the user packets. The method for reducing an average delay time in a packet transfer network according to claim 10. 12. The storage unit of the edge node includes: a first information table for determining a service class corresponding to attribute information given to a user packet and predetermined for a user; And a second information table storing a frame length corresponding to a traffic load situation in the lightwave network, and referring to the first table and the second table of the storage unit, 12. The method according to claim 11, wherein a frame length of the light adaptation frame accommodating a packet is determined. 13. The packet transfer network according to claim 11, wherein the edge node selects a frame length for each priority indicated by an attribute of a user packet received from a subscriber network. Method. 14. The method according to claim 11, wherein the edge node selects a frame length according to a load condition in the lightwave network. 15. Synchronous digital hierarchy (SDH)
Utilizes wavelength division multiplexing (WDM) technology in the lower layer of
An edge node that accommodates a subscriber network and includes a lightwave adaptation layer that builds a lightwave adaptation frame for converting a user packet from a subscriber and transfers the lightwave network into a lightwave network, and a core node that performs relay transfer processing in the core network In the lightwave network including: the core node observes the number of transfer packets passing per unit time, outputs the observation information, and the central monitoring device receives the observation information output by the core node, A status determination is performed, and a traffic load status determination result is output to the edge node. The edge node accommodates a user packet from the subscriber network based on the determination result received from the central monitoring device. The feature is to determine the frame length of the light wave adaptation frame Average delay time shortening method in a packet transfer network for. 16. A subscriber access network accommodating a subscriber network, including a subscriber access device for receiving a user packet from the subscriber and transferring the packet to a subsequent data communication network, and connecting to the subscriber access network. And a packet transfer network including a data transfer network including a data transfer device that performs a packet relay transfer process. The packet transfer network converts a received user packet and transfers the converted user packet to the packet transfer network. An adaptation layer constructed as a new frame, wherein the subscriber access device selects a frame length of the frame for transfer of the packet transfer network according to an attribute of a user packet received from the subscriber network; An average delay time shortening method in a packet transfer network, wherein the transfer is performed to a data transfer network. 17. A subscriber access network accommodating a subscriber network, receiving a user packet from the subscriber, and transferring the packet to a subsequent data communication network, and a connection to the subscriber access network. A packet transfer network including a data transfer network including a data transfer device that performs a relay transfer process of the packet, wherein the packet transfer network converts a received user packet into a new packet and transfers the converted packet to the packet transfer network. The subscriber access device selects a frame length of the frame for transfer of the packet transfer network according to an attribute of a user packet received from the subscriber network, and 17. The method according to claim 16, wherein the data is transferred to a transfer network. 18. A communication network comprising a subscriber access network accommodating a subscriber network, receiving a user packet from the subscriber, and transferring the packet to a subsequent data communication network, wherein the subscriber access network comprises: From the attribute information of the input user packet, extract a plurality of user packets belonging to the same attribute, attribute information on the extracted user packet, and
A frame length for transfer in the data communication network is selected based on a load state of the data communication network, and the selected user packet having the selected frame length and including the extracted user packet as payload information is transferred to a subsequent data communication network. A method for reducing the average delay time in a packet transfer network, characterized in that the method is constructed as a frame.