JPH11196101A - Packet transfer condition monitoring device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a packet transfer condition monitoring device capable of realizing the efficient operation of a network. SOLUTION: This device is provided with a monitoring/analysis part 100 for monitoring and analyzing the transfer conditions of a packet transferred to a transmission destination by a routing processing in a network layer, a tendency prediction part 101 for predicting the tendency of the transfer of the packet based on a monitored result by the monitoring/analysis part 100 and a path setting condition judgement part 102 for judging a condition for shifting to a hardware processing below a data link layer from a result predicted by the tendency prediction part 101.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a packet transfer status monitoring device.

[0002]

BACKGROUND OF THE INVENTION Since the transmission of data between computers over a telecommunications network, any communication network has served as an infrastructure for the computer. The Internet, which is made by connecting computers and LANs in various parts of the world in a network, is now becoming an indispensable part of industry in the world.

[0003] The main services of the original Internet are:
It used to be electronic mail, file transfer (FTP), remote login (telnet), etc.
The number of computers connected to the Internet has exploded since the wide-area information system in the hypertext format called the orldWide Web) became mainstream. In addition, with the increase in speed and performance of CPUs, the amount and size of data handled on computers has also increased, and this trend is expected to continue with the increase in multimedia data in the future.

[0004] As a result, the problem of the capacity of communication networks has been greatly highlighted. The problem of the capacity of the communication network includes a problem of performance of a server in each region, a problem of a capacity of a communication line itself, and a problem of IP forwarding of a router.

[0005] TCP / I represented by the Internet
A network in which a plurality of networks are connected using the P is connected by a device called a router that relays packets at a network layer level. The router reads the sending and receiving addresses of the packets to be relayed, and relays only the packets that are permitted to be relayed. However, this process is often performed by software, and as the number of packets to be forwarded increases, the load on the software processing of the router increases. The overload causes delay in packet transmission and packet loss due to buffer overflow. This is one of the major causes of the above-mentioned problems related to IP forwarding in the router.

As a means for solving this, the present applicant has proposed
He was involved in the development of SR (Cell Switch Router). This is a next-generation high-speed router using label switching technology. Here, the CSR will be described as an example. The conventional router performs processing at the network layer level by the CPU every time all incoming packets are transferred, and transfers the packets. This network layer level process determines where incoming packets are to be transferred. When transferring large data, the number of packets increases, and the same process is repeated for each individual packet. become.

[0007] On the other hand, the CSR processes the packet received first at the network layer, and when the information of the header of the packet corresponds to the preset information (cut-through trigger information), Subsequently, the incoming packet is transferred (cut-through transfer) only by processing below the data link layer. As a result, the processing at the network layer level that places a heavy load on the router is reduced.

Further, the processing in the network layer is performed by software, but the processing in the data link layer and below can be realized by hardware, so that the processing can be performed at higher speed with less delay. Further, since an ATM switch on an ATM network, which has already been introduced as a backbone communication network, can be used below the data link layer, migration from an existing environment is easy and cost performance is high.

[0009] Due to these characteristics, CSR has attracted attention as a next-generation router on a growing TCP / IP network.

[0010]

As described above, the basis for determining whether or not to perform cut-through transfer is the cut-through trigger information. In the current specification, this cut-through trigger information must be set in advance by an administrator. However, when the administrator sets the cut-through trigger information in response to a request from the user, it is difficult to determine that the request really reduces the load on the router and is effective for the network.

Further, after setting the cut-through trigger information, confirming whether or not the cut-through trigger information is working effectively can be a heavy burden on the administrator. For example, if the setting is such that cut-through is performed even for the transfer of a small number of packets, the delay caused by the process of shifting to the cut-through mode may be larger.

[0012] In order to design effective cut-through trigger information, the actual IP packet transfer situation must be carefully considered, but it is very difficult, and this requires the use of label switching technology including CSR. Router has become a problem.

The present invention has been made in view of such a problem, and an object of the present invention is to enable setting of an effective cut-through path by monitoring a packet transfer state in a router. Another object of the present invention is to provide a packet transfer status monitoring device capable of realizing efficient network operation.

[0014]

To achieve the above object, a packet transfer state monitoring apparatus according to a first aspect of the present invention comprises:
Monitoring means for monitoring a transfer status of a packet transferred to a destination by a routing process in a network layer; prediction means for predicting a tendency of packet transfer based on a monitoring result by the monitoring means; Determining means for determining a condition for shifting to hardware processing at the data link layer and below from the result predicted by the above.

The packet transfer status monitoring device according to a second aspect of the present invention is the packet transfer status monitoring device according to the first aspect of the present invention, wherein the packet transfer status monitoring device performs hardware processing at a data link layer or lower according to the condition determined by the determination means. And setting means for automatically setting trigger information for instructing the transition of.

Further, a packet transfer state monitoring device according to a third aspect of the present invention is a monitoring means for monitoring a transfer status of a packet transferred by hardware processing at a data link layer or lower, and based on a monitoring result by the monitoring means. Determining means for determining whether or not the hardware processing is effectively performed, and automatically releasing or resetting trigger information for instructing the hardware processing according to the determination result by the determining means. Release resetting means.

[0017]

DESCRIPTION OF THE PREFERRED EMBODIMENTS First, an outline of the present embodiment will be described. The present embodiment intends to make the cut-through trigger information that has been fixedly set conventionally more effective. In order to design effective cut-through trigger information, it is first necessary to monitor and analyze the actual packet transfer status. There are two points to be monitored. Considering IP packets, monitoring of protocol information and port numbers of IP packets to be transferred, and IP
This is the monitoring of the source IP address of the packet and the destination IP address.

These are currently two modes of cut-through mode in label switching technology, namely, flow mode.
It is realized by the driven mode and the topology-driven mode. In the flow-driven mode, the protocol information and the port information of the header of the IP packet are shifted to the cut-through mode. And a mode for shifting from the destination IP address to cut-through.

FIGS. 1A and 1B are diagrams for explaining the outline of the packet transfer state monitoring apparatus according to the present embodiment.
In the present embodiment, with the above two modes in mind, first, the monitoring / analyzing unit 100 shown in FIG. Next, based on the monitoring and analysis results, the trend prediction unit 101
Medium and long term IP packet transfer trends are predicted. Then, the path setting condition determination unit 102 determines a condition under which the cut-through mode is considered to function effectively, that is, a flow and a topology from the predicted result. And
The automatic setting / cancellation unit 103 uses this condition to automatically set the cut-through trigger information in accordance with a temporal change, and cancels the setting if it is found that the setting is invalid. Perform

Also, the following processing is performed for the one for which cut-through trigger information has already been set. In other words, in FIG. 1B, the use status monitoring unit 200 monitors the use status of the cut-through trigger information for which the cut-through trigger information has already been set, and the validity determination unit 201 makes the setting function effectively. It is determined whether or not. And, for those that are determined not to be working effectively,
The resetting unit 202 automatically cancels the cut-through trigger information, and resets it if a more effective condition is found.

As described above, by employing the effective cut-through trigger information, the load on the router using the label switching technique and the network including the router can be reduced, and the high-speed packet transfer and the network Efficient operation can be realized.

Hereinafter, the above outline will be described in more detail with reference to the drawings. FIG. 2 is an example of a network incorporating a CSR. Host A (10) -Host B in the figure
Consider the operations on the CSR 13 when file transfer is performed between (11), between the host B (11) and the host C (12), and between the host A (10) and the host C (12).
The IP packet that has entered the CSR 13 is sent to the IP processing unit (IP controller) 13 of the CSR 13 as indicated by the arrow in FIG.
-1 to be processed at the network layer level (IP processing) and transferred to the next hop (CSR 15). At the time of the IP processing at this time, the protocol information, the port number, the source IP address, and the destination IP address are read from the header of the IP packet, collated with the cut-through trigger information, and entered into the cut-through mode. A transition is determined. In the case of CSR, when transitioning to the cut-through mode, the neighboring C
It communicates with the SR and negotiates the cut-through path.

FIG. 4 shows the movement of the CSR when a cut-through trigger is applied. The protocol information, the port number, the source IP address, and the destination IP address of the IP packet which do not shift to the cut-through mode are monitored for medium and long term, and the transfer status of the IP packet as shown in FIG. Create data.

In the example shown in FIG. 7, the number of transferred IP packets is totaled every three hours. The part indicated by ／/ x means a specific month / day, and △ / △ means that the next day has changed. When data is collected for one week, it is assumed that the following tendency is predicted from the aggregated data. 1. TCP between Host A and Host B almost every day from 18:00 to 3:00
A large number of / 23 packets are transferred, but only a small number of packets are transferred at other times. 2. Between host A and host C, many UDP / 300 packets are transferred only on weekdays. On holidays, no UDP / 300 packet transfer can be seen. 3. TCP / between host B and host C between 9:00 and 21:00 on weekdays
21 packets are transferred in large numbers. On holidays and holidays, a very small amount of packets are transferred.

In this case, it can be determined that the following conditions are effective for setting the cut-through trigger information. 1. TCP / 2 only between host A and host B from 18:00 to 3:00
Setting of cut-through trigger information of 3 is valid. 2. Setting of fixed cut-through trigger information of UDP / 300 between host A and host C is valid. 3. TC only between 9:00 and 21:00 on weekdays between host B and host C
P / 21 cut-through trigger information setting is valid.

In order to determine the valid cut-through trigger information as described above, the number of transfer packets per unit time during which the cut-through mode is valid must be used as a reference. The number of transfer packets per unit time as a reference is calculated in consideration of the performance of the process of shifting the CSR to the cut-through mode.

The recommended cut-through determined in this manner
The trigger information is automatically set according to the temporal change.

Further, the network usage status changes daily, and it is possible that the recommended cut-through trigger information once determined becomes invalid. Therefore, by monitoring whether or not the operated cut-through mode is working effectively, invalid cut-through trigger information is canceled.

However, at the time of cut-through, unlike the hop-by-hop transfer shown in FIG. 3, when ATM transfer is used for layers 2 and below, IP processing is not performed as shown in FIG. Not done. Therefore, instead of monitoring the information of the IP header, the number of ATM cells is set to AT
Monitoring is performed for each of the M port, VPI, and VCI values.

At this time, the reference is the number of cells per unit time. If the number of cells per unit time in the monitoring result does not exceed the reference number, it is determined that cut-through is not working effectively. Then, the set invalid cut-through trigger information is automatically released. Also, if the cut-through transfer has already been performed at the time of this release processing, the cut-through transfer and the hop-off from the cut-through transfer as shown in FIG.
Force a switch to bi-hop forwarding.

In the above-described embodiment, the transfer state of the packet is monitored over time and cut-through and
Although the trigger information is set, the present invention is not limited to this. For example, the cut-through trigger information may be set according to the geographical condition where the router is located or the priority of the packet to be transferred. Good.

[0032]

According to the present invention, the router using the label switching technology itself determines and automatically sets conditions for shifting to efficient cut-through, and also monitors the effectiveness of subsequent cut-through transfer. As a result, efficient network operation can be realized and the load on the network administrator can be reduced.

[Brief description of the drawings]

FIG. 1 is a diagram schematically illustrating a packet transfer status monitoring device according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating an example of a network incorporating a CSR.

FIG. 3 is a diagram showing a flow of an IP packet at the time of hop-by-hop transfer.

FIG. 4 CSR when cut-through trigger is applied
FIG. 6 is a diagram showing the movement of the image.

FIG. 5 is a diagram showing the movement of the CSR during the cut-through operation.

FIG. 6 is a diagram illustrating a transition operation from cut-through transfer to hop-by-hop transfer.

FIG. 7 is a diagram illustrating an example of a result of monitoring a transfer state of an IP packet in CSR.

[Explanation of symbols]

 Reference Signs List 100: monitoring / analysis unit, 101: trend prediction unit, 102: path setting condition judgment unit, 103: automatic setting / cancellation unit, 200: usage status monitoring unit, 201: validity judgment unit, 202: cancellation / resetting unit .

Claims (3)

[Claims] 1. A monitoring means for monitoring a transfer status of a packet transferred to a destination by a routing process in a network layer, and a prediction means for predicting a tendency of packet transfer based on a monitoring result by the monitoring means. A packet transfer state monitoring device comprising: a determination unit configured to determine a condition for shifting to hardware processing at a data link layer or lower based on a result predicted by the prediction unit. 2. The apparatus according to claim 1, further comprising setting means for automatically setting trigger information for instructing a transition to hardware processing at a data link layer or lower in accordance with the condition determined by said determining means. Packet transfer status monitoring device. 3. A monitoring means for monitoring a transfer status of a packet transferred by hardware processing at a data link layer or lower, and based on a monitoring result by the monitoring means, whether the hardware processing is effectively performed. Determining means for determining whether or not the trigger information for instructing the hardware processing is automatically released or reset in accordance with a result of the determination by the determining means. Packet transfer status monitoring device.