JP2000341328A - Data repeater - Google Patents

Abstract

PROBLEM TO BE SOLVED: To shorten the processing time by excluding redundancy of copy processing. SOLUTION: A cache memory 200a of a low-order layer stores received data in the unit of packets (1), a table generating means CCTC generates a cache management table CCT denoting cross reference between data copied to a buffer 300a of an application layer from the low-order layer with a packet on the cache memory 200a and stores it to a buffer 300 (2), after data processing, a difference information generating means DIFC uses the cross reference stored in the table CCT to generate information specifying a packet needing revision and different information DIF having a revised position and revised contents in the packet (3), an update section UPD of the low-order layer uses the difference information to update the packet stored in the cache memory 200a and the result of update is used for transmission data (4).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for relaying communication between terminals (relay apparatus). The present invention relates to a relay device that performs predetermined data processing in a certain application layer and transmits data according to a processing result to the outside via a lower layer.

[0002]

2. Description of the Related Art (a) Conventional network environment In a recent network environment, a relay device RT such as a router is installed at a boundary portion between the Internet and an intranet as shown in FIG. In many cases, transmission data is created by performing the processing described above, and the transmission data is transmitted downstream. For example, the application of the relay device RT performs a security check on the data arriving from the terminal T1 via the connection CN1, and locates the transmission data updated or newly created by the check processing at the next position on the communication path. It transmits to terminal T2 via connection CN2. In such a communication mode, the relay device is required to have higher processing performance in order to realize comfortable communication.

The relay device RT having a software function shown in FIG. 30 is composed of hardware HW and software, and the software further includes an OS (hereinafter referred to as a kernel KNL) for mainly performing basic control such as protocol processing, and a kernel KNL. And an application APL that runs on In such a configuration, the application APL
Usually cannot directly access the memory on the kernel KNL or hardware HW. For this reason, in order for the application APL to perform software processing such as security check in the relay device RT, it is necessary to execute the process shown in FIG.
As shown in (i) to (ii) of FIG.
L and hardware HW) to copy the received data from the memory area to the memory area accessible by the application APL, and after processing the data, to the lower layer memory area as shown in (iii) to (iv). It is common to perform an operation. However, in such a method, the copying operation is required many times, and the relay process takes time.
Therefore, there is a demand for speeding up of the relay process.

FIG. 30 shows a configuration in which the relay device RT performs both data transmission / reception processing and security check processing.
The check processing may be configured to be performed by another device (application server). FIG. 31 shows a relay device
FIG. 31 is a configuration diagram in a case where an application server ASV is provided separately from the RT, and the same components as those in FIG. 30 are denoted by the same reference numerals. The difference is that an application server ASV is provided, and data is transmitted and received between the relay device RT and the application server ASV via the connection CN10. The application of the relay device RT simply performs data transmission / reception processing, and connects the received data through the route shown in the figure.
Sends to the application server ASV via CN10, performs a check process in the application server ASV, sends the updated or newly created transmission data by the check process to the relay device RT, and the relay device RT relays the data and communicates Connection CN2 to terminal T2 located next on the route
To send over. In the configuration of FIG. 31, the same copy operation as in FIG. 30 is required in the node, and (i) to (i)
Copying between nodes as shown in i) is required, and the time during which data stays in the relay system becomes longer, so that a higher speed of the relay processing is demanded.

(B) Internet Protocol Hierarchical Structure As shown in FIG. 32, the Internet protocol is composed of a four-layer model of a link layer, an Internet layer, a transport layer, and an application layer from the lower layer. The hardware HW of the relay device RT and the server application ASV in FIGS. 30 and 31 is a portion corresponding to the link layer, the kernel KNL is a portion corresponding to the Internet layer / transport layer performing IP / TCP processing, and the application APL is This is the part corresponding to the application layer.

[0006] The link layer is a layer responsible for performing highly reliable data transfer between adjacent nodes, and differs for each access technology of a transmission medium used in a local network. For example, in LAN (Local Area Network), Ethernet.Token rin
g, a layer that performs access protocol processing such as FDDI, and a layer that performs protocol processing such as X.25 and ATM in a wide area network. The unit of the message handled in the link layer is called a frame or a packet. As shown in FIG. 33, a link header is arranged at the head of the frame and a trailer is arranged at the end.

The Internet layer is a layer that provides a service for transmitting a message from a transmitting terminal to a destination terminal in cooperation with a relay device such as a router. Internet Protocol (IP: Inte
rnet Protocol) has been standardized. The unit of message handled at the Internet layer is called a datagram,
As shown in FIG. 33, an IP header is arranged at the head. The transport layer is a layer that provides a service for sending a message from a sending application process to a receiving application process. As a transport protocol, a TCP protocol (TCP: transmission control protocol) that guarantees high reliability of communication is standardized. The unit of the message handled in the transport layer is called a segment, and a TCP header is arranged at the head of the segment as shown in FIG.

[0008] The application layer is a layer that provides services for each user application. For example,
Functions for performing file transfer, remote login, e-mail transfer, etc. are defined, and protocols according to each are standardized. Nodes such as terminals and relay devices add a header to the data passed from the application layer into segments by the transport layer, add a header to the segments into datagrams in the Internet layer, and add a header to the datagram in the link layer. And a trailer are added to form a frame and transmitted to a physical medium.

[0009] (c) Control of relay device [0010] There is an application generally called a proxy application, which performs data processing such as security check in the relay device and relays data to the next terminal on the communication path. I do. This is, as shown in FIG.
This is an application that checks data passing through the relay device (terminal T3) existing at the boundary of No. 2 and further performs relay processing. In FIG. 34, connection CN1, CN
2 is between terminal T1 and terminal T3, terminal T3 and terminal T2, respectively.
It is a connection established between

In the following, the operation of a conventional proxy application will be described and its problems will be clarified. In addition,
Buffers that can be directly accessed by the application ALP, the kernel KNL, and the hardware HW are respectively described as “application buffer or buffer A”, “kernel buffer or buffer K”, and “hardware buffer or buffer H”. In order to distinguish between “data” handled by the application APL and “received data” stored in the cache memory, the former is referred to as “data” and the latter as “packet”. Further, it is assumed that the IP header and the TCP header have a well-known configuration shown in FIGS. 6-bit control flag in TCP header
s (CF) is a field indicating the type of segment, in which various flags are defined, and the SYN flag indicates that this is the first segment when a connection is opened.

(C-1) Packet Reception First, the procedure performed by the terminal T3 in FIG. 34 from the reception of the packet by the hardware HW to the reception of the data by the proxy application APL is described in FIG. The description will be made according to the flow. (1) Receiving a packet by the hardware HW The hardware HW on the terminal T3 fetches the packet if the destination hardware unique address (MAC address) indicated by the link header of the packet matches its own MAC address. , At address 1 on the buffer H.

(2) Receiving a packet by the kernel KNL A processing unit (described by a driver) on the kernel KNL for accessing the hardware layer copies a packet on the address 1 to an address 2 on the buffer K, and 1 is deleted. (3) Connection establishment procedure by kernel KNL The kernel KNL has the destination IP address of terminal T3 from the IP header of the received packet, and the upper layer protocol is TC
When recognizing that the protocol is the P protocol, it refers to the SYN flag of the TCP protocol header and checks whether the connection CN1 has been established.

If the SYN flag is set, the kernel KNL
Is a request to open a new connection by combining the source IP address al, destination IP address p1, source PORT number a2, and destination PORT number p2 (al, pl, a2, p2) in the IP / TCP protocol header of the received packet. Recognize as a packet. Hereinafter, a connection having this combination is referred to as a connection CN1, and a combination of the above four entries (al, pl, a2, p2)
Is referred to as a socket pair. Next, the kernel KNL completes the connection CN1 with the terminal T1 by a connection establishment procedure in TCP and sends a message to the proxy application to notify the completion of the connection. Thereafter, the kernel KNL relates to the interface CN1 (referred to as socket 1) between the kernel KNL and the application APL with respect to the connection CN1, and receives a reception socket buffer RSB1, a transmission socket buffer SSB1, and an update waiting buffer R as buffers dependent on the socket 1.
NB1 is newly created.

Each socket buffer has a queue structure. The receiving socket buffer RSB1 can store a plurality of addresses of received packets on the buffer K in order from the head of the queue. Are stored in a plurality of transmission packet addresses. By the above procedure, socket pair (al, pl,
The procedure for establishing the connection CN1 when the TCP connection corresponding to a2, p2) has not been established is completed. In the above, the processing is performed when a connection opening request is received from the terminal T1, but when opening a connection from the terminal T3 to the terminal T1, the socket pair becomes (p1, a1, p2, a2)
A request to open a new connection is transmitted from the terminal T3 to the terminal 1, and thereafter, the connection CN is established between the terminal T1 and the terminal T1 by the connection establishment procedure in the TCP.
Complete 1

(4) Storing the Received Packet Address in the Receive Socket Buffer Thereafter, the terminal T3 sets the socket pair (a
With reference to (l, pl, a2, p2), the address 2 of the buffer K storing the received packet is queued in the queue (matrix) of the receiving socket buffer RSB1 corresponding to the socket pair. "Save packet in buffer queue"
This means that the head address of the buffer K storing the packet is connected to a queue (matrix) of the receiving socket buffer and stored.

(5) Received Packet Existence Message A When the sum of the packet lengths stored in the receive socket buffer RSB1 exceeds a certain upper limit, the kernel KNL notifies the proxy application APL of the existence of the received packet by a message “message”. "A" is transmitted via the socket 1. (6) Data read message D When message A is received, the proxy application
The APL transmits a “data read message D” to the socket 1 with the address 4 of the buffer A as the “read destination address” and the data length as the “data length” with arguments as arguments.

(7) Writing to the application buffer A by the kernel KNL The kernel KN receiving the “data read message D”
L reads data from the storage area (buffer K) indicated by address 2 stored at the head of the queue of the reception socket buffer RSB1 and writes it as one piece of data at address 4 of buffer A specified by message D excluding the protocol header. (Copy of data). If there are a plurality of packets to be written, the kernel KNL reads data from the storage area (buffer K) indicated by the next address from the reception socket buffer RSB1, and writes the data continuously to the buffer A except for the protocol header. Thereafter, the kernel KNL deletes the read address from the reception socket buffer RSB1 and deletes the packet stored at address 2 of the kernel buffer K. With the above mechanism, the application can receive data from the connection 1.

(C-2) Proxy application APL that has completed creation of transmission data and reception of transmission data
If there is a change to the received data during the data processing such as security check, re-create the data and transfer the created data to the proxy application APL.
And the transmission data from If necessary, the proxy application APL creates new transmission data. Below, after the proxy application APL creates transmission data, the hardware HW
Until the transmission data is transmitted to the terminal TN2 via the terminal T.
3 shows the procedure to be performed. FIG. 37 shows a packet transmission flow,
FIG. 38 is an explanatory diagram of creation and transmission of transmission data. (8) Creation of transmission data The proxy application APL creates transmission data at the address 10 in the buffer A by the transmission data creation means.

(9) Data write message P If there is data to be transferred to the terminal T2, the proxy application APL determines that the connection data from the terminal T3 to the terminal T2 has not been established yet, The connection CN2 is established in accordance with (3).
Thereafter, the application APL sends the address 1 of the buffer A to the interface dedicated to transmission / reception of the connection CN2 (hereinafter referred to as socket 2) as the “head address”.
A "data write message P" is sent to the kernel KNL with "0" and "transmission data length" on the address 10 as "data length" as arguments. (10) Copying Transmission Data to Kernel Upon receiving the “data write message P”, the kernel KNL copies the transmission data at the address 10 to the address 20 in the buffer K. . (11) Next, the kernel KNL sends the send socket buffer SSB2
Queue the address 20 of the buffer K in the queue. The transmission socket buffer queue is a queue for storing a head address of transmission data, and performs protocol processing on transmission data sequentially from the head of the queue.

(12) Remaining Protocol Processing When creating the TCP / IP protocol header, the transmission socket 2 uniquely determines the identification number of the TCP header for each packet, and sets the SEQ number of the TCP header to the It is determined based on the byte number of the data transmitted from CN2, and further, the TCP header checksum and the IP header checksum are calculated by a calculation formula defined by the TCP / IP protocol. The created TCP / IP protocol header is added to the packet of the address 20. If the terminal T3
If the hardware layer has a protocol processing unit such as a checksum calculation unit in the hardware layer, the TCP header checksum and the IP header checksum are replaced with the result of the operation using the protocol processing unit. Alternatively, it is also possible to transfer the packet to the hardware layer without changing these fields, and then create a packet to be finally transmitted by a protocol processing unit that calculates a checksum thereafter.

(13) Protocol Processing of Transmission Socket After completing the protocol processing, the kernel KNL retransmits the contents of the transmission socket buffer SSB2 to the retransmission socket buffer SSB.
Copy to 2 '. Where retransmit socket buffer SSB
2 'is a mechanism for realizing a retransmission mechanism, which is a standard function of the TCP protocol. Retransmission will be described later. (14) MAC address determination and transmission After all TCP / IP protocol headers have been created,
Kernel KNL is a routing table (routing table)
Determines the destination MAC address based on the
Is set as the source MAC address (14-1). And a link header (MAC header) containing these MAC addresses
Is added to the datagram, and the packet is stored in the address 13 of the hardware buffer H (14-2). To the terminal TN2 (14-3).

(C-3) Operation of Socket in TCP / IP An operation example using a socket in conventional TCP / IP is as follows. At the time of data transmission from the terminal T3, the newly created data at the address 10 of the application buffer A is copied to the address 20 in the kernel buffer K,
The address 20 is stored in the transmission socket buffer SSB2. The kernel KNL performs TCP / IP protocol processing on the data at the address 20 stored in the transmission socket buffer SSB2. When the creation of the TCP / IP protocol header is completed, the kernel KNL sends the address 20 to the retransmit socket buffer SSB2 '.
While transmitting the packet stored in the address 20 in the step (14), and deleting the address 20 from the transmission socket buffer SSB2.

Thereafter, if a packet A having a SEQ number greater than the SEQ number of the packet S at address 20 stored in the retransmission socket buffer SSB2 'is received from the terminal T2 within a predetermined time, the packet S is received by the terminal T2. And deletes it from the retransmitted socket buffer SSB2 '.
However, if the packet A is not received by a certain waiting time, the packet is taken out from the address 20 stored in the retransmission socket buffer SSB2 'and this packet is retransmitted. With the above mechanisms (c-1) to (c-2), the conventional proxy application checks the security of data transmitted between terminals and implements the relay of the communication. .

[0024]

As described above, conventionally,
In order for the application of the relay device located at the boundary of the network to perform processing such as security check on the received data and transmit the updated or newly created transmission data to the next terminal on the communication path, 1) Copy the received data from the memory area of the kernel KNL and hardware HW to the accessible memory area of the application APL.
It is necessary to perform redundant operations such as sequentially copying and transmitting to the memory area of the KNL and the hardware HW. Also, in the configuration of FIG. 31, inter-node copy is required in addition to intra-node copy. As described above, conventionally, the copying operation is required many times, it takes a long time to relay, and hinders high-speed data transmission.

Accordingly, a first object of the present invention is to be located at the boundary of a network, perform application processing such as security check on received data, and transmit the data to the next terminal on a communication path. Thus, in a relay apparatus that relays data, high-speed communication can be performed while eliminating redundancy of copy processing as much as possible. A second object of the present invention is to significantly reduce the redundancy of copy processing between nodes in a relay apparatus in which an application for performing data processing such as security check is provided in an application node different from the relay node. This is to enable high-speed communication.

[0026]

Means for Solving the Problems Data to be transmitted from the relay device to the outside may be compared with received data before processing when only a specific portion of the field is changed, or may or may not be relayed without being changed. In most cases, only the selection is necessary, and the portion that needs to be changed is usually very small in the entire received data. The present invention has been made focusing on this point.

According to the first relay apparatus of the present invention, the cache memory of the lower layer stores the received data in packet units, and the table creating means stores the data copied from the lower layer to the buffer of the application layer with the cache. A cache management table indicating the correspondence with the packet on the memory is created and stored in the buffer, and after the data processing, the difference information creating unit uses the correspondence held in the table to identify the packet that needs to be changed. The difference information having the information to be specified, the changed portion in the packet, and the changed content is created, and the update unit of the lower layer updates the packet stored in the cache memory using the difference information, and transmits the update result to the transmission data. And In this way, it is only necessary to copy the difference information from the upper layer to the lower layer, so that the processing time can be reduced by eliminating the redundancy of the copy processing.

According to the second relay device of the present invention,
The lower layer cache memory stores the received data in packet units, and the table creating means creates a cache management table indicating the correspondence between the data copied to the application layer buffer and the packets in the cache memory, and creates a lower layer cache management table. After the data processing by the application, the difference information creating means stores the information specifying the data that needs to be changed among the data copied to the application buffer, and the difference information including the changed portion and the changed content of the data. The update unit of the lower layer specifies a packet that needs to be changed using the difference information and the correspondence held in the cache management table, and updates the packet to be transmission data. In this way, it is only necessary to copy the difference information from the upper layer to the lower layer, so that the processing time can be reduced by eliminating the redundancy of the copy processing.

In the above first and second relay devices, when the difference information creating means determines not to transmit the data in the cache memory to the outside, the difference information includes a deletion instruction, and the updating means performs the deletion. The data in the cache memory is deleted according to the instruction. In this way, the processing time can be shortened simply by instructing the lower layer to delete from the upper layer. Further, when no change is made to the data in the cache memory, the difference information creating means includes a no-change instruction in the difference information, and the updating unit uses the data in the cache memory as the transmission data as it is in accordance with the no-change instruction. In this way, the processing time can be shortened simply by instructing the lower layer to perform no change from the upper layer. Further, the difference information creating means causes a new instruction to be included in the difference information when newly creating transmission data, and the updating means sets the data included in the difference information according to the new instruction as new transmission data. In this way, in any case of data change, deletion, no change, and new creation, the lower layer can be instructed by the difference information of the same format. When the number of changed portions for the same packet is larger than the set number, or when the total sum of the changed data lengths for the same packet is longer than the set length, the application creates new transmission data. In this way, the amount of difference information can be reduced, and the processing time can be reduced.

According to the third relay device of the present invention,
In the first node, the cache memory stores the received data in packet units, and the table creating unit creates a cache management table indicating the correspondence between the copy destination data in the application buffer of the first node and the packet in the cache memory. , The communication unit transmits the received data and the cache management table to the application of the second node, and in the second node, the table correction unit compares the correspondence of the cache management table with the copy destination data on the application buffer in the second node. After correcting the correspondence to the packet in the cache memory, and after performing the data processing, the difference information creating unit specifies the packet in the cache memory that needs to be changed using the correspondence held in the corrected cache management table. Changes in information and packets The communication means transmits the difference information to the first node, and the updating unit of the first node updates the packet stored in the cache memory using the difference information. The update result is used as transmission data. In this way, it is only necessary to copy the difference information from the second node to the first node, so that the processing time can be reduced by eliminating the redundancy of the copy processing.

According to the fourth relay device of the present invention,
In the first node, the cache memory stores the received data in packet units, and the table creating unit creates a cache management table indicating the correspondence between the copy destination data in the application buffer of the first node and the packet in the cache memory. , The communication means transmits the received data to the application of the second node via a predetermined connection, and the conversion table creating means generates the conversion table indicating the correspondence between the head address of the data copied to the application buffer and the connection identifier. In the second node, after the data processing, the difference information creating unit creates information specifying data in the second application buffer that needs to be changed, and difference information having a changed portion and changed content in the data, The communication means sends the difference information to the first node. The difference information correcting means of the first node corrects the difference information using the data head address of the second application buffer included in the difference information and the data head address of the first application buffer obtained from the conversion table, and updates the difference information. Specifies a packet that needs to be changed using the corrected difference information and the correspondence held in the cache management table, and updates the packet to be transmission data. According to the above, only the data is transferred from the first node to the second node, and only the difference information is transferred from the second node to the first node. Can be shortened.

According to the fifth relay device of the present invention,
In the first node, the cache memory stores the received data in packet units, and the table creating unit creates a cache management table indicating the correspondence between the copy destination data in the application buffer of the first node and the packet in the cache memory. The communication means transmits the received data to the application of the second node via a predetermined connection, and the identification table creating means determines the identification indicating the correspondence between the head address of the data copied to the application buffer and the identifier of the data. A table is created, and in the second node, the difference information creating unit creates information specifying data in the second application buffer that needs to be changed, and difference information having a changed location and changed content in the data, and a communication unit. Is the difference information
Transmitted to the node, the difference information correcting means of the first node corrects the difference information using the data head address of the second application buffer included in the difference information and the data head address of the first application buffer obtained from the identification table, The updating unit specifies a packet that needs to be changed using the corrected difference information and the correspondence held in the cache management table, and updates the packet as transmission data. According to the above, only the data is transferred from the first node to the second node, and only the difference information is transferred from the second node to the first node. Can be shortened.

[0033]

DESCRIPTION OF THE PREFERRED EMBODIMENTS (A) Outline of the Present Invention (1) First Application Processing of the Present Invention FIG. 1 is a schematic explanatory diagram of a first application processing of the present invention, and FIG. FIG. 4B is an explanatory diagram of the flow of application processing, and FIG. 4B is an explanatory diagram of a cache management table. Although not shown, the relay device includes a hardware HW, a kernel KNL, and an application APL.

(1-1) Configuration The cache means CSH provided in the kernel KNL copies a packet received from the hardware HW to the application buffer 300a and copies the received packet (copy source data) in the kernel buffer 200. Cache memory
Store (cache) in 200a. The cache management table creation means CCTC provided in the cache means CSH1,
A table (cache management table) for associating the copy destination data on the application buffer 300a with the copy source data (packet) on the cache memory 200a.
CCT) is created and stored in the application buffer 300. Cache management table CCT, when there are multiple packets PT ₁ ~PT _n 1 data, manages to indicate the correspondence between the data and the packet (b) for each packet. When the difference data is created later, the received packet to be changed can be specified by referring to the cache management table CCT.

Means D for creating difference information of application APL
The IFC specifies a packet including a changed portion in the data by referring to the cache management table CCT, and creates difference information DIF including the packet specifying information, the changed portion of the packet, and the changed content. The updating means UPD provided in the kernel KNL replaces, in the packet on the cache memory 200a specified by the difference information DIF created by the difference information creating means DIFC, the changed portion specified by the information with the changed contents. Creates transmission data.

(1-2) Operation Packet received externally by the hardware HW of the relay device
The PTi is copied by the kernel KNL to the application buffer 300a. At that time, the cache means CSH stores the received packet PTi in the cache memory 200a in the kernel buffer.
Save. Also, the cache management table creation means CC
The TC creates a cache management table CCT in the application buffer 300 that stores a correspondence between the stored received packet PTi (i = 1 to n) and the data copied on the application buffer 300a. Then, application A
The PL performs processing such as security check on the received data (PT1 to PTn), and the difference information creating means DIFC stores the packet PTi of the cache memory 200a according to the data portion that needs to be changed based on the check result in the cache management table C.
Identify by referring to CT. Then, the difference information creation means DI
FC1 includes information for identifying the packet PTi to be changed,
A difference information DIF describing a changed portion of the packet and the contents of the change is created in the memory 300b, and the difference information DIF is created.
The IF is copied to the memory 200b accessible by the updating means UPD.

When receiving the difference information DIF from the application APL, the updating means UPD specifies the packet PTi in the cache memory 200a to be changed by the difference information DIF, and then specifies the changed portion in the packet PTi. Then, the data of the changed portion is replaced with the changed contents, and this is used as transmission data. Finally, the hardware HW transmits this packet to the outside, so that communication between terminals can be relayed.

(2) Second Application Process of the Present Invention FIG. 2 is a schematic explanatory diagram of a second application process of the present invention, wherein FIG. FIG. 4 is an explanatory diagram of a cache management table. The second application processing differs from the application processing of FIG.
T means to update kernel buffer 200 accessible by UPD
It is a point created within. The hardware HW of the relay apparatus copies the packet received from the outside to the PTi, and the kernel KNL copies the packet to the application buffer 300a. At that time, the cache means CSH is the cache memory 2 in the kernel buffer.
The received packet PTi is stored in 00a. The cache management table creating means CCTC also stores the received packet
(i = 1 to n) and a cache management table for storing the correspondence between the data copied on the application buffer 300
CCT update means Kernel buffer 20 accessible by UPD
Create within 0.

The application APL performs a security check and the like on the received data (PT1 to PTn). Then, based on the check result, the difference information creating means DIFC creates difference information DIF describing the data specifying information, the changed portion related to the data, and the changed content, and stores the difference information DIF in the memory 3.
00b, and copies the difference information DIF to the memory 200b accessible by the updating means UPD. Application APL
UPD receives the difference information DIF from the
The packet including the portion corresponding to the changed portion of the data to be changed by the DIF is identified with reference to the cache management table CCT, and the information of the changed portion in the packet is identified by the difference information DI.
Replace with the changed contents of F, and use this as the transmission data. Finally, the hardware HW transmits this packet to the outside, so that communication between terminals can be relayed.

(3) Third Application Processing of the Present Invention FIG. 3 is a schematic explanatory diagram of the third application processing of the present invention, and shows a flow of application processing of the relay device. The third application process is different from the first application process in FIG. 1 in that the kernel KNL can directly access the memory 100 on the hardware HW. 100 is shown to be included. The cache unit CSH copies the packet PTi received by the hardware HW to the application buffer 300a, and stores (caches) the received packet PTi (copy source data) in the cache memory 100a in the hardware buffer.

Operation Packets received externally by the hardware HW of the relay device
The PTi is copied to the application buffer 300 by the kernel KNL. At that time, the cache unit CSH stores the received packet PTi in the cache memory 100a in the hardware buffer 100. Also, the cache management table creation means CCTC includes the stored received packet PTi (i = 1 to n),
A cache management table CCT indicating the correspondence with the data copied on the application buffer 300 is created and stored in the application buffer 300. The application APL performs a security check and the like on the received data (PT1 to PTn), and the difference information creating unit DIFC stores a packet PTi including a portion to be changed in the data based on the check result in the cache management table CCT.
Specify using. Then, the difference information creation means DIFC1
Creates, in the memory 300b, information for specifying the packet PTi to be changed, and a difference information DIF describing the changed portion of the packet and the details of the change, and the difference information DIF
To the memory 200b accessible by the updating means UPD1.

When receiving the difference information DIF from the application APL, the updating means UPD1 specifies the packet PTi in the cache memory 100a to be changed by the difference information DIF, and then specifies the changed portion in the packet PTi. Then, the data of the changed part is replaced with the changed contents, and this is used as transmission data. Finally, the hardware HW transmits this packet to the outside, so that communication between terminals can be relayed. The updating means UPD may be provided in the hardware HW.

(4) Fourth Application Processing of the Present Invention FIG. 4 is a schematic explanatory diagram of the fourth application processing of the present invention, and shows a flow of application processing of the relay device. The fourth application processing differs from the third application processing in FIG. 3 in that the cache management table CCT is created in the hardware buffer 100 accessible by the updating means UPD. The packet HTi received from the outside by the hardware HW of the relay device is
NL copies to application buffer 300. At that time, the cache unit CSH stores the received packet PTi in the cache memory 100a in the hardware buffer 100. Also, the cache management table creation means CCTC,
A cache management table CCT indicating the correspondence between the stored received packet PTi (i = 1 to n) and the data copied on the application buffer 300 is created and stored in the hardware buffer 100.

The application APL performs a security check and the like on the received data (PT1 to PTn). Then, based on the check result, the difference information creating means DIFC creates difference information DIF describing the data specifying information, the changed portion related to the data, and the changed content, and stores the difference information DIF in the memory 3.
00b, and copies the difference information DIF to the memory 200b accessible by the updating means UPD. Application APL
UPD receives the difference information DIF from the
The packet including the portion corresponding to the changed portion of the data to be changed by the DIF is identified with reference to the cache management table CCT, and the information of the changed portion in the packet is identified by the difference information DI.
Replace with the changed contents of F, and use this as the transmission data. Finally, the hardware HW transmits this packet to the outside, so that communication between terminals can be relayed.

(5) Fifth Application Processing of the Present Invention FIG. 5 is a schematic explanatory diagram of the fifth application processing of the present invention. When a received packet is partially changed and transmitted, unnecessary received packets are specified. The following describes application processing / update processing when the packet is not transmitted, when the received packet is transmitted without being changed, and when a new packet is created and transmitted.
(5-1) Configuration After security check processing,
Based on the check result, it is determined whether there is a received packet that needs to be partially changed, there is an unnecessary received packet that needs to be deleted, there is a received packet that is transmitted without any change, and there is a new packet. If there is a packet to be partially changed, the difference information creating means DIFC creates difference information DIF for specifying the packet, the changed portion in the packet, and the changed content, and stores the packet in the memory 200 in the kernel buffer.
Copy to b.

If there is an unnecessary packet, the deletion notifying means DLTN creates deletion information DLT for specifying the unnecessary packet and copies it to the memory 200b in the kernel buffer. If there is a packet that does not need to be changed, the non-change notifying means NCGN creates non-change information NCG specifying the unchanged packet and copies it to the memory 200b in the kernel buffer. If there is a new packet newly created by the application APL, the new data notification means NDTN
Copy the NPK to memory 200b in the kernel buffer.
The update and transmission processing of the transmission data based on the difference information DIF has already been described with reference to FIGS. However, it is assumed that the received data including the plurality of received packets PT1 to PTn is copied to the application buffer 300a and stored in the cache memory 200a according to the procedure described in FIG.

(5-2) Data deletion After the application APL performs security check processing,
Based on the check result, it is determined whether there is a received packet that needs to be partially changed, there is an unnecessary received packet that needs to be deleted, there is a received packet that is transmitted without any change, and there is a new packet. If an unnecessary packet exists in the received data, the application APL refers to the cache management table CCT (FIG. 1) to specify the unnecessary packet on the cache memory 200a. After that, the deletion notifying means DLTN creates deletion information DLT for deleting the unnecessary packet and copies it to the memory 200b in the kernel buffer. Upon receiving the deletion information from the application APL, the updating means UPD deletes the packet on the cache memory 200a specified by the deletion information DLT.
As a result, the deleted packet is not transmitted. The above is a case where some packets have been deleted. However, all the packets constituting the received data may be deleted. In such a case, there is no data to be transmitted, and the relay process ends immediately.

(5-3) Transmission of Unchanged Packet After the application APL performs security check processing,
Based on the check result, it is determined whether there is a received packet that needs to be partially changed, there is an unnecessary received packet that needs to be deleted, there is a received packet that is transmitted without any change, and there is a new packet. If there is a packet to be transmitted without any change, the application APL refers to the cache management table CCT (FIG. 1) to specify the unchanged packet on the cache memory 200a. After that, the non-change notifying means NCGN creates non-change information NCG for notifying an unchanged packet and copies it to the memory 200b in the kernel buffer. Upon receiving the non-change information NCG from the application APL, the updating means UPD does not change the packet on the cache memory 200a indicated by the no-change information NCG. As a result, the packet is transmitted as received.

In the above description, some of the packets are unchanged. However, there are cases where all the packets constituting the received data are unchanged. In such a case, after the check processing, the received data is immediately transmitted as it is. That is, if the application APL determines that the received data is not changed after the check processing, the application APL specifies all the packets in the cache memory 200a corresponding to the data by referring to the cache management table CCT. After that, no change notification means N
The CGN creates unchanged information NCG for transmitting a specific packet without change, and copies it to the memory 200b on the kernel buffer. The updating means UPD receiving the information from the application APL specifies a packet which is the target of the unchanged information NCG, and uses it as transmission data. Finally, the hardware HW can relay the communication between the terminals by transmitting the unchanged packet to the outside.

(5-4) Transmission of New Data After the security check processing, the application APL
Based on the check result, it is determined whether there is a received packet that needs to be partially changed, there is an unnecessary received packet that needs to be deleted, there is a received packet that is transmitted without any change, and there is a new packet. If the application APL determines that a new packet should be created, it creates a new packet NPK. After that, the new data notification means NDTN copies the new packet NPK to the memory 200b in the kernel buffer. Application APL
, The updating means UPD adds the new packet to other packets to be transmitted and sets it as transmission data. After that, the hardware HW can transmit the transmission data to the outside, thereby relaying the communication between the terminals.

In the above description, a part of the transmission data is added as a new packet. However, the entire transmission data may be transmitted as a new packet. In such a case, after the check processing, the application APL creates new transmission data. The new data notification means NDTN copies the new transmission data to the memory 200b in the kernel buffer. When new transmission data is received from the application, the updating means UPD recognizes that the data is newly created data, and uses this as transmission data. After that, the hardware HW sends this new data to the outside,
It becomes possible to relay communication between terminals.

(6) Sixth Application Process of the Present Invention FIG. 6 is a schematic explanatory diagram of the sixth application process of the present invention. The sixth application processing differs from the first application processing of FIG.
A different point is that the APL is provided with new data creation determining means NDCD for determining whether to create new data. Based on the result of the application processing on the received data stored in the application buffer 300a, the new data creation determining means NDCD grasps the number of changed portions of the received data or the received packet, and compares the number of changed portions with the set number X. . If the number of changed portions is smaller than the set number X, the difference information creation means DIFC is instructed to create difference information, and
If the number of changed parts is larger than the set number X, the new data creation unit NDTG is instructed to create new transmission data instead of the difference information. FIG. 7A shows an example of determining whether to create difference information or new transmission data based on the size of the changed portion and the set number X. FIG. 7B shows the total data length and the set length of the changed portion. This is an example in which it is determined which of differential information and new transmission data is to be created according to the size of the data.

The kernel KNL copies the packet PT received from outside by the hardware HW of the relay device into the application buffer 300a. At that time, the cache means CSH
Saves the received packet PTi in the cache memory 200a in the kernel buffer. Also, the cache management table creation means CCTC stores the received packet PTi (i = 1 to n)
A cache management table CCT is created in the application buffer 300 to store the correspondence between the data and the data copied on the application buffer 300a. The application APL performs a process such as a security check on the received data (PT1 to PTn), and the new data creation unit NDC calculates the number of changed portions of the received data (or received packet) based on the check result. If the number of locations is smaller than the set number X, an instruction to create difference information is issued,
If the number of changed parts is larger than the set number X, it instructs to newly create transmission data instead of the difference information.

If the number of changed portions is smaller than the set number X, the difference information creating means DIFC uses the method described in FIG.
Is created, and the difference information DIF is copied to the memory 200b.
Upon receiving the difference information DIF from the application APL, the updating means UPD1 specifies the packet PTi on the cache memory 200a to be changed by the difference information DIF, and then
The changed portion in the packet PTi is specified, the data of the changed portion is replaced with the changed content, and the data is used as transmission data. Finally, the hardware HW transmits this packet to the outside, so that communication between terminals can be relayed. On the other hand, if the number of changed portions is larger than the set number X, the new data creation unit NDTG creates new data and copies the new data to the memory 200c in the kernel buffer. When new transmission data is received from the application, the updating means UPD recognizes that the data is newly created data, and uses this as transmission data. After that, the hardware HW sends this new data to the outside

In the above, it is determined whether to create difference information or new transmission data based on the magnitude of the number of changed portions and the set number X. However, as shown in FIG. It is also possible to determine whether to create difference information or new transmission data based on the length Y. For example, in the case of FIG. 7B, the sum of the changed bit lengths A + B + C
It is determined which of the difference information and the new transmission data is to be created, based on the magnitude of the setting length Y. The above is the case where the application of the relay apparatus performs processing such as security check and transmits data to the outside, and is an example applicable to the configuration shown in FIG. By the way, as shown in FIG.
Node different from the relay device (application server)
There is a configuration in which the above application performs a process such as a security check and transmits it to the outside via a relay device.
The following is an example applicable to the configuration as shown in FIG.

(7) Seventh Application Process of the Present Invention FIG. 8 is a schematic explanatory diagram of the seventh application process of the present invention. The seventh application processing is different from the first application processing in FIG. 1 in that a relay node 1 and a node 2 including a data processing application are separately provided, and the applications on both nodes transfer mutually necessary information. That is, the application of the node 2 performs predetermined processing on the data received from the outside by the node 1. The communication unit with table CMTB copies the received data and the cache management table CCT from the application on the node 1 to the application on the node 2. The communication means CM transfers data between the application on the node 1 and the application on the node 2. The communication means CM and CMTB can be realized by, for example, an existing transfer protocol between different nodes using a connection.

The table correction means TBCR corrects the cache management table. The cache management table CCT 'transferred from the node 1 via the table-attached communication means CMTB first shows the correspondence between the data of the application buffer 300a on the node 1 and the received packet on the cache memory 200a of the node 1. I have. Therefore, the table modifying means TBCR modifies the table CCT 'so as to indicate the correspondence between the data in the application buffer 300a' on the node 2 and the received packet on the cache memory 200a of the node 1. Hereinafter, the processing flow of the application will be described. The kernel KNL of the relay node 1 copies the packet PTi received from outside by the hardware HW to the application buffer 300a. At that time, the cache means CSH stores the received packet PTi in the cache memory 200a in the kernel buffer. Also, the cache management table creation means CCTC uses the stored received packet
A cache management table CCT indicating the correspondence between the PTi (i = 1 to n) and the data copied on the application buffer 300a is created in the application buffer 300.

Next, the application of the node 1 copies the received data to the application of the node 2 via the communication means 11 with a table. Further, the application of the node 1 copies the created cache management table CCT (intermediate table in the figure) together with the data to the application of the node 2 and, after copying, copies the intermediate table CC
Delete T. Next, the application of the node 2 uses the table modifying means TBCR to execute the cache management table CC.
T is corrected so that the corrected cache management table CCT 'indicates the correspondence between the data of the application buffer 300a' on the node 2 and the received packet on the cache memory 200a of the node 1. Further, the application of the node 2 performs a process such as a security check on the received data.

After the check processing, the difference information creating means DIFC of the application sends a packet in the cache memory 200a corresponding to the data portion that needs to be changed based on the check result.
The PTi is specified with reference to the cache management table CCT '. Next, information specifying the packet that needs to be changed, a difference portion DIF describing the changed portion of the packet, and the contents of the change are created in the memory 300b ′, and the difference information DIF is stored in the application buffer 300b on the node 1 through the communication means CM. Copy the difference information DIF. The application on the node 1 that has received the difference information DIF copies the difference information DIF to the memory 200b accessible by the updating means UPD. Update means U
PD receives difference information DIF from application,
The cache memory 2 to be changed by the difference information DIF
00a to identify the packet PTi, and then
Is changed, the data of the changed part is replaced with the changed contents, and this is used as transmission data. Finally, the hardware HW transmits this packet to the outside, so that communication between terminals can be relayed.

(8) Outline of Eighth Application Process of the Present Invention FIG. 9 is a schematic diagram of the eighth application process of the present invention, and FIG. 10 is an explanatory diagram of address conversion. The difference between the eighth application process and the seventh application process (FIG. 8) is that the received packet PTi (i =
1 to n) and a cache management table for specifying the correspondence between the data copied on the application buffer 300a
A point that the CCT is not transferred to the application of the node 2 but is held in the kernel buffer 200 of the node 1. A conversion table CVTB is provided in the application buffer 300 of the node 1, and the conversion table CVTB includes communication means (referred to as connection X). And application buffer 300a
The difference information creation means DIFC specifies the portion of the received data that needs to be changed, using the address of the address buffer 300a ', based on the check result of the application. ,
Creating the difference information DIF having the changed portion, the changed content, etc.,

The difference information correcting means DICR is provided in the application of the node 1, and the difference information DIF is calculated using the data head address P of the application buffer 300a and the data head address Q of the application buffer 300a 'stored in the conversion table CVTB. The updating means UPD of the node 1 refers to the cache management table CCT for the received packet including the portion corresponding to the changed portion included in the difference information DIF in that the address indicating the changed portion included in the difference information DIF is converted into the address of the application buffer 300a. Then, the information of the changed portion in the packet is replaced with the changed content of the difference information DIF, and this is used as transmission data.

In FIG. 9, the communication means CMN 1
This is a means for the application of the node 1 to transfer data to the application of (1). This communication means CM
N1 can be realized by the existing transfer protocol between different nodes using the connection X. Further, a conversion table CBTB is created that associates the connection X used in the transfer with the storage position of the transferred data (the data start address P of the application buffer 300a). The communication means CMN2 is means for the application of the node 2 to transfer the difference information DIF to the application of the node 1, and the communication means CMN2 can be realized by a different node transfer protocol using the connection X.

Hereinafter, the flow of the processing of the application will be described. When the hardware HW of the node 1 receives a packet from outside, the kernel KNL copies the received packet to the application buffer 300a. At that time, the cache means CSH is the cache memory 2 in the kernel buffer.
The received packet PTi is stored in 00a. The cache management table creating means CCTC also stores the received packet
(i = 1 to n) and a cache management table CC for storing the correspondence between the data copied on the application buffer 300a
T means to update kernel buffer 200 accessible by UPD
Create within. Next, the application of the node 1 copies data (received data) to the application buffer 300a 'of the node 2 by the communication means CMN1. Further, the application communicates the data on the node 1 (the head address P of the application buffer 300a) with the communication means CMN1.
Creates a conversion table CVTB that specifies the correspondence relationship with the connection X used in the memory 300 accessible by the application of the node 1.

The application of the node 2 performs processing such as security check on the received data.
Then, based on the check result, the difference information creating means DIFC creates difference information DIF describing the data specifying information, the changed portion of the data (the changed portion is specified using the address of the address buffer 300a '), and the changed content. And stored in the memory 300b ', and the connection X (communication means CMN2)
And stores the difference information DIF in the application buffer 300b 'of the node 1. The difference information correction means DICR provided in the application of the node 1 is based on the conversion table CVTB.
Start address P of buffer 300a held in
And the data head address Q of the buffer 300a ′,
The address indicating the changed part included in the difference information DIF is converted into the address of the buffer 300a, and the converted difference information is copied to the memory 200b accessible by the updating means UPD.

The updating means UPD of the node 1 specifies the received packet including the portion corresponding to the changed portion included in the difference information DIF by referring to the cache management table CCT, and determines the information of the changed portion in the packet by the difference information DIF. Is replaced with the changed contents, and this is used as transmission data. Finally, the hardware HW transmits this packet to the outside, so that communication between terminals can be relayed.

FIG. 10 is a diagram for explaining the operation of the conversion table CVTB. Data is transferred from the application of the node 1 to the application of the node 2 via the connection X. At the time of this data transfer, the data start address in the application buffer 300a of the transfer source (node 1) is P, and the data start address of the application buffer 300a 'in the transfer destination (node 2) is Q. In this case, the difference information creating means DI on the node 2
The head memory address of the data targeted by the FC is Q, and the updating means UPD cannot correctly update the data without any change. That is, data based on the head memory address Q must be converted into data based on the head memory address P. Therefore, a conversion table CVTB that associates the connection X, which is the communication means CMN1, with the data head address P of the application buffer 300a is created when data is transferred from the node 1 to the node 2.

Then, when the difference information DIF is received from the “connection X” at the time of transferring the difference information from the node 2 to the node 1, the difference information correcting means DICR reads the head address P from the conversion table CVTB, and Is changed from Q to P. As a result, the difference information DIF becomes data based on the head memory address P. Therefore, the updating means UPD can perform the subsequent packet update processing in the same manner as when the difference information is created in the application layer of the node 1 thereafter.

(9) Outline of Ninth Application Process of the Present Invention FIG. 11A is a schematic diagram of a ninth application process of the present invention. This ninth application process is shown in FIG.
The difference from the eighth application process is that an identification table RGTB is provided instead of the conversion table CVTB.
The purpose of the identification table RGTB is the same as that of the conversion table CVTB of FIG. It is assumed that data is transferred from the application of the node 1 to the application of the node 2 via the connection X. At the time of this data transfer, the application buffer 30 of the transfer source (node 1)
It is assumed that the data head address at 0a is P and the data head address of the application buffer 300a 'at the transfer destination (node 2) is Q.

In such a case, the head memory address of the data targeted by the difference information creating means DIFC on the node 2 is Q, and the updating means UPD cannot correctly update the data as it is. That is, data based on the head memory address Q must be converted into data based on the head memory address P. Therefore, a unique identification number X is assigned to each transfer source data, and the start address P of the data is assigned.
Is associated with the identification number X (see FIG. 11).
(B)) is created at the time of data transfer from the node 1 to the node 2, and the communication means CMN1 transmits the identification number X together with the data. The difference information creation means DIFC creates difference information DIF including the identification number X. In other words, the difference information creating means DIFC generates the difference information DIF having the data specifying information, the changed portion relating to the data (the changed portion is specified based on the start address Q of the address buffer 300a '), the changed contents, and the identification number X. create.

When the difference information is transferred from the node 2 to the node 1, the difference information correcting means DICR reads the head address P from the identification table RGTB by referring to the identification number X included in the difference information DIF. Then, the head memory address of the data targeted for the difference information is changed from Q to P, and the “identification number” is deleted from the difference information. As a result, the difference information DIF becomes data based on the head memory address P. Therefore, the updating means UPD can perform the subsequent data update processing in the same manner as when the difference information is created in the application layer of the node 1 thereafter.

(B) First Embodiment FIG. 12 is a conceptual diagram of a relay according to the present invention, and is a terminal corresponding to the present invention installed in the middle of a route connecting terminals T1 and T2 that perform communication using the TCP / IP protocol. FIG. 9 is a network configuration diagram illustrating a relay operation of (relay device) T3. The terminal T3 receives the data addressed to the terminal T3 from the connection CN1, relays the data to the terminal T2 via the connection CN2, and also relays the reverse route, thereby enabling communication between the terminal T1 and the terminal T2. And the application to relay is running on terminal T3. In the following, the operation performed by the terminal T3
Details of (packet reception, creation of differential data structure, packet transmission) will be described. The first embodiment is an embodiment that implements the processing described with reference to FIGS. 1 and 5 to 7.

(A) Reception of Packet FIG. 13 is a flow chart of packet reception. The proxy application APL is composed of a plurality of received packets after the hardware HW of the terminal T3 receives the packet PTi from the connection CN1. Shows the procedure performed by the terminal 3 until the terminal 3 receives the. (1) Receiving a packet by hardware The hardware HW on the terminal T3 fetches the packet if the MAC address indicated by the link header of the packet matches its own MAC address,
It is stored at address 1 (addresses 1-1 and 1-2 in the figure) on 0. (2) Packet reception by kernel KNL The processing unit (driver) on the kernel KNL for accessing the hardware layer transfers the packet on the address 1 to the address 2 on the buffer K (addresses 2-1 and 2- in the figure). Copy to 2) and delete the packet on address 1.

(3) Procedure for Opening Connection by Kernel KNL When the kernel KNL recognizes from the IP protocol header that the destination IP address is that of the terminal T3 and that the upper-level protocol is the TCP protocol, Refer to the SYN flag of the header, and connect CN1
Check if has been established. If the SYN flag is set, the kernel KNL uses a combination of the source IP address al, destination IP address p1, source PORT number a2, and destination PORT number p2 (al, pl, a2, p2) is recognized as a new connection establishment request packet. Hereinafter, a connection having this combination is referred to as a connection CN1, and a combination (al, pl, a2, p2) of the above four entries is referred to as a socket pair. Thereafter, the kernel KNL completes the connection CN1 with the terminal T1 by a connection establishment procedure in TCP and sends a message to the proxy application APL to notify the completion of the connection. Next, the kernel KNL is connected to CN1
As for the interface between the kernel KNL and the application APL (referred to as socket 1), and as a buffer subordinate to the socket 1, the reception socket buffer RSB
1, send socket buffer SSB1, update wait buffer RNB1
Is newly created.

Each of the buffers created above has a queue structure, and the receiving socket buffer RSB1 sequentially receives the addresses 2-1 and 2-1 of the received packet on the buffer K from the head of the queue.
2-2... Can be stored, and the transmission socket buffer SSB1 stores a plurality of addresses of transmission packets on the buffer K. Hereinafter, a description of a buffer queue is used for each of the above buffers. Through the above procedure, the procedure for establishing the connection CN1 when the TCP connection corresponding to the socket pair (al, pl, a2, p2) has not been established is completed. In the above, processing was performed when a connection establishment request was received from the terminal T1,
When a connection is established from the terminal T3 to the terminal T1, the socket pair becomes (p1, a1, p2, a2), and a request for opening a new connection is transmitted from the terminal T3 to the terminal 1. The connection CN1 is completed with the terminal T1 by a connection establishment procedure in TCP.

(4) Storing the Received Packet Address in the Receive Socket Buffer Queue Thereafter, the terminal T3 sets the socket pair (a
l, pl, a2, p2), the addresses 2-1-2, 2 of the kernel buffer 200 storing the received packet in the queue (matrix) of the receive socket buffer RSB1 corresponding to the socket pair.
Queue the -2, ... (5) Received packet existence message A When the sum of the packet lengths stored in the receive socket buffer RSB1 exceeds a certain upper limit, the kernel KNL sends a “message A”, which is a message notifying the proxy application APL of the existence of the received packet. Send to the socket 1.

(6) Packet number request message B When the proxy application APL previously associated with the socket 1 receives the message A from the socket 1,
A memory area for reading data is prepared at address 4 in the application buffer 300, and a “packet number request message B” for asking the number of packets to be read now is transmitted to the socket 1 using the data length as an argument. (7) Packet Number Response Message C The kernel KNL sends a “packet number response message C” that responds to the “packet number request message B” with N as the number of corresponding packets.

(8) In response to the message C, the proxy application APL allocates a memory area necessary for creating a cache management table CCT describing the correspondence between data and a packet group at the address 5 in the application buffer 300.
To secure. The cache management table CCT is a table describing the correspondence between the data stored in the application buffer 300 and the received packet group PTi (i = 1 to N), and one is created for one piece of received data. As an example, if one received data in the application buffer 300 and two packets in the kernel buffer 200 are associated with each other as shown in FIG. 14, the cache management table CCT representing this association is as shown in FIG. Become.

Generally, the cache management table CCT is created for each of the N packets PTi (i = 1 to N), and stores the “head address” of the storage area in the kernel buffer 200 storing each packet. ”, Its“ data length ”, and a table group having a“ next table pointer ”indicating the head memory address of the next packet table (hereinafter, table CCT-1, table CCT-2,..., Table CCT-N Notation), and the application buffer to read data from
A table having a “start address”, a “data length”, a “number of packets” to be read, and a “next table pointer” indicating the start address of the memory of the table CCT-1 (hereinafter, table CCT-h and To be described). In the example of FIG. 14, the cache management table CCT includes a table CCT-h, a table CCT-1, and a table CCT-2.

(8) Size of Cache Management Table CCT Since the size of the cache management table CCT depends on the number of packets constituting received data, the proxy application APL recognizes the number of packets N in the “packet number response message C”. Then, secure the memory and create the cache management table CCT (the contents of each field are blank except some). For example, in an environment in which each field length is set to 4 bytes in advance as shown in FIG. 15, if the result of the “packet number response message C” is N = 2, the cache management table CCT is the table CCT-h and the table CCT. -1, the table CCT-2, the total length of which is 40 bytes, the proxy application APL secures a 40-byte memory area at the address 5 in the application buffer 300, and stores a partial field of the cache management table CCT. (Such as the start address in table CCT-h).

(9) Data Read Message D Thereafter, the proxy application APL sends the socket 1 a read destination address (= address 4), a data length, a table CCT read destination address (= address 5), and the number of read packets. (= 2) is transmitted as a data read message D, respectively.

(10) Writing to Application Buffer by Kernel KNL Kernel KN that Received “Data Read Message D”
L is written as one piece of data at address 4 excluding the protocol header from the data (packet) stored at address 2-1 queued at the head of the receive socket buffer RSB1, and at the same time, In the “Start address” field and the “Data length” field,
Write the address 2-1 and the data length (= 1024) of the packet to be read, respectively. When there are a plurality of packets to be written (N = 2 in the example of FIG. 14), every time the i-th packet is written, a predetermined value is written to each field of the table CCT-i, and After writing i in the “number of packets” field and writing all data, NULL is written in the “next table pointer” of the table CCT-N to complete the creation of the cache management table CCT.

Thereafter, the kernel KNL deletes the read address from the reception socket buffer RSB1, but does not delete the data on the address 2 of the kernel buffer 200 itself. However, there is a case where the data (packet) on the address 2 is to be deleted after the data is copied to the address 4 due to the exhaustion of the kernel buffer 200 or the like. In such a case, the kernel KNL determines whether or not to delete the data.
Notify that PTi will not be saved. Thus, the application APL can recognize whether or not the packet PTi is stored in the kernel buffer 200 by referring to the start address of the table CCT-i.

With the above mechanism, the cache means, that is, the packet copied to the application buffer 300 is stored in the address 2 (cache memory) of the kernel buffer, and the correspondence between the packet in the cache memory and the data in the application is stored. Management table CCT that manages
And a cache means can be realized. The cache management table CCT can be stored in a lower layer (kernel layer), which will be described in a second embodiment.

(B) Creation of a difference data structure FIG. 16 is an explanatory diagram of a difference data structure creation process of an application. The proxy application APL that has completed the data reception performs processing such as security check, and if the received data changes in the process,
Create the difference information DIF (change location, contents of change, etc.) and input it to the kernel KNL. Creates data and sends it out from the hardware HW. The application APL creates new data if necessary. Below, kernel buffer 20
The processing of the proxy application APL in an environment where the received packets PTi (i = 1 to 2) are stored in the cache memory (addresses 2-1 and 2-2) in 0 will be described.

(11) Creation of Difference Data by Proxy Application When the proxy application APL completes data reception and completes storage of the cache management table CCT, it performs data processing such as security check based on the data. . In the course of this data processing, if a changed portion occurs in the data, the application difference information creating means DIFC refers to the cache management table CCT and has the changed position of the packet corresponding to the changed portion, the data content after the change, and the like. Create difference information DIF. Then, (the packet that needs to be changed, the difference data structure DCB such as the change position in the packet) is stored at the address 10 in the application buffer 300, and the changed content (data after rewriting) RRD is stored at the address 11.

If the “head address” of the table CCT-i (i = 1, 2,...) Is NULL, the new data creation determining unit NDCD regards that the received packet is not stored in the kernel. Then, it instructs the new data creation means NDTG to create new data. Further, the new data creation determining unit NDCD instructs the new data creation means NDTG to create new data even when the number of changed portions of the received data is large or the number of changed bits is long. Note that the new data creation determination unit NDCD determines the difference information creation means when it determines not to create new data.
Instruct DIFC to create difference information

(12) Difference Data Structure Format As shown in FIG. 17, the format of the difference data structure is a memory address D1, a new data flag D2, a deleted data flag D3, and an update area in the packet cache memory that need to be changed. Information (changed part) D4, data byte position D5 and data bit position D before update
6.Data byte length D7 and data bit length D8 before update, data byte length D9 and data bit length D10 after update, additional / unchanged information D11 on the same packet, pointer indicating the location to store the updated data contents D1
2 as a member. The data position is specified in bit units in the form of “byte position D5” + “bit position D6”. Of the above members, "Bit position D6", "Bit length D
8, D10, "additional change presence / absence information D11,""new data flag D2," and "deletion data flag D3" have a length of 1 byte, and the remaining members have a length of 4 bytes. Details of each structure member will be described later (13) “Difference information creation means” and (14)
This will be described in "New data creation means".

(13) Difference Information Creation Means As shown in FIG. 14, the 2048-byte data received by the proxy application APL from the kernel KNL is stored in the address 4 of the application buffer 300, and two corresponding receptions are performed. Packet PT1, PT2 (N = 2)
Are stored at addresses 2-1 and 2-2 in the kernel buffer 200 (cache memory), and a cache management table CCT describing the correspondence is stored at address 5 of the application buffer 300 as shown in FIG. It is assumed that -When making changes only to the protocol header Receive from the connection CN1 of the socket pair (a1, p1, a2, p2) and send the application buffer 3 via the kernel KNL.
Connection CN2 of socket pair (b1, q1, b2, q2) without changing the contents of the data copied to address 4 of 00
When transmitting data to the cache, the difference information creating means DIFC creates difference information so as to change only the TCP / IP protocol header of the received packets PT1 and PT2 in the cache memory.

That is, if there is no need to change the data content, the difference information creating means DIFC performs the processing described below to obtain the T
Create difference information DIF (difference data structure DCB and rewritten data RRD) for changing the socket pair of the CP / IP protocol header. That is, the following processes 1 to 3 are performed on the packet PT1.
The same processing is performed for the second packet PT2.

First, the difference information creating means DIFC
The cache management table CCT of FIG. 2 recognizes that the packets PT1 and PT2 corresponding to the data of the address 4 to be changed exist in the cache memory (addresses 2-1 and 2-2), Address as "D1"
Set 2-1 (D1 = address 2-1). Next, the difference information creating means DIFC secures 30 bytes as a memory area for creating the difference data structure DCB based on the cache management table CCT in FIG. Is set.
The “update area information D4” is information that preliminarily divides a packet into several areas and specifies which of the areas is to be changed. For example, if the IP header is Area 1, the TCP header is Area 2,..., And the data part used by the application is Area 50, the update area information D4 = 1 indicates that the changed area is the IP header. I can tell you. If there is no change area, D4 = NULL.

Next, the byte position D5 and the bit position D6 of the data before update are respectively indicated by the transmission source in the IP header.
The byte length D7 and the bit length D8 of the data before updating are specified from the position of the IP address a1, and the byte length D7 and the bit length D8 of the data before the update are each specified by the length of the source IP address a1. Figure 3 shows the IP protocol header
5 has a structure shown in (a), source IP address, from the head 12 bytes + 0-th bit to the beginning 32-bit (= 4
Byte + 0 bit). Therefore, D5 = 12 bytes, D6 = 0
Bit, D7 = 4 bytes, D8 = 0 bit. Since the byte length D9 and the bit length D10 of the updated data are 32 bits ( 4 bytes + 0 bits) before and after the update, respectively, D9 = 4 bytes and D10 = 0 bits in order.

Next, the difference information creating means DIFC newly secures a memory area for storing the changed contents (the rewritten data RRD) at the address 11 in the application buffer 300, and stores the “rewritten data RRD” here. save. In this example, the updated source IP address "b1" is written. Further, the pointer D12 of the difference data structure DCB is set to address 11. In this example, since all socket pairs are changed with respect to the same received packet, "additional change presence / absence information D11 regarding the same packet" is set to 1 in order to declare that there is a change in data other than the source IP address. Set to.
If there are no further changes, D1 = 0. The “new data flag D2” indicates that the changed content (the rewritten data RRD) indicated by the pointer D12 is the new data creation means ND.
Specifies whether the data is new data created by the TG. If it is new data, D2 = 1; if not, D2
2 = 0. Here, D2 = 0 because it is not new data.

The “deletion data flag D3” sets D3 = 1 when the proxy application APL decides to discard the received packet, and sets D3 = 0 in other cases. Here, D3 = 0 is set because data is not discarded. Hereinafter, a difference data structure DCB is similarly created for the remaining socket pairs (q1, q2, q3). "Information D1 of additional changes
Set "1" to 0. As described above, the creation of the difference information DIF instructing the change of the IP / TCP header, that is, the creation of the difference data structure DCB and the rewritten data RRD (= b1, q1, b2, q2) is completed.
However, although not adopted in the present embodiment, a configuration may be adopted in which the kernel KNL that recognizes that the packet is relayed to the connection CN2 at the time of transmission changes the protocol header.

When a change is made to the data contents When a part of the data stored at the address 4 of the application buffer 300, for example, 24 bytes from the 2024th byte of the data is deleted, the difference data structure DCB is as follows.
Create However, the socket pair is changed in the same manner as described above. Proxy application difference information creation means DIFC
Recognizes that the packet corresponding to the data at the address 4 to be changed exists in the cache memory (addresses 2-1 to 2-2), and further recognizes that the 2024th byte is the packet PT2. Address 2-2 is set as “packet memory address D1”. The area 50 is set as the update area information D4 (D5 = 50), and the change of the data part is declared.

The byte position D5 and the bit position D6 of the data before update are set to the byte position D5 and the bit position D6 of the data before update at the 1000th byte (1000 bytes + 0 bits) which is the change position in the packet PT2. The byte length D9 and the bit length D10 of the updated data are specified in (byte + 0 bit), respectively (0 byte + 0 bit). That is, D5 = 1000 bytes, D6 = 0 bits, D7 = 24 bytes, D8 = 0 bits, D9 = 0 bytes, and D10 = 0 bits. In this example, to delete the specified area, the “data RRD after rewriting” is not created, and the pointer D12 of the difference data structure DCB is set to NULL. In this example, in order to change all the socket pairs with respect to the same received packet, the “additional change presence / absence information D11 regarding the same packet” is set to 1. D2 = 0 for the "new data flag". For the “deletion data flag”, D3 = 0. As described above, the creation of the difference data structure DCB when a part of the data content is deleted is completed.

When the data is not changed When the data stored in the address 4 of the application buffer 300 is transmitted without being changed, the following difference data structure DCB is created. Here, "unchanged" means that the socket pair itself is not changed at all. The packet PT1 is subjected to the following differential data structure creation processing (1), and after the processing is completed, the address 2-
The same processing is performed for the second packet PT2. Proxy application difference information creation means DIFC
Indicates that the packets PT1 and PT2 corresponding to the data at the address 4 to be transmitted unchanged are stored in the cache memory.
Recognizing that it exists at (addresses 2-1 and 2-2), address 2-1 is set as "packet memory address D1". "Update area information D4" is set to 0, and it is declared that there is no change area. The byte position D5 and bit position D6 of the data before update, the byte length D7 and bit length D8 of the data before update, and the byte length D9 and bit length D10 of the data after update are all set to NULL.

In this example, since the received packet PT1 is unchanged, the “rewritten data RRD” is not created.
Therefore, the pointer D12 indicating the data position after rewriting is NULL
Set L. In this example, since no other change is made, “additional change presence / absence information D11 on the same packet” is set to 0. "New data flag" is set to D2 = 0. The “deletion data flag” is set to D3 = 0. By creating the difference data structure DCB as described above, it is possible to notify the kernel KNL that the packet PT1 on the address 2-1 is transmitted without being changed. Similarly, a difference data structure is created for the packet PT2 on the address 2-2.

When Discarding Data When deleting the entire data stored at the address 4 of the application buffer 300, a difference data structure DCB is created as follows. Note that the following differential data structure creation processing is performed for the packet PT1, and after the processing is completed, the same processing is performed for the packet PT2 of the address 2-2. Proxy application difference information creation means DIFC
Recognizes that the packets PT1 and PT2 corresponding to the data at the address 4 to be discarded exist in the cache memory (addresses 2-1 and 2-2), and recognizes that the packet memory address D
Address 2-1 is set as "1". "Update area information D4" is set to 0, and it is declared that there is no change area. The byte position D5 and bit position D6 of the data before update, the byte length D7 and bit length D8 of the data before update, and the byte length D9 and bit length D10 of the data after update are all set to NULL.

In this example, since only the received packet is deleted, the “rewritten data RRD” is not created. Therefore, the pointer D12 indicating the data position after rewriting is NULL.
Set. In this example, since no other change is made, “additional change presence / absence information D11 on the same packet” is set to 0. "New data flag" is set to D2 = 0. "Delete data flag" is D3 = to discard the packet.
Set to 1. By creating the differential data structure DCB as described above, it is possible to notify the kernel KNL that the packet PT1 on the address 2-1 is discarded and not transmitted. Similarly, a difference data structure is created for the packet PT2 on the address 2-2.

(14) New Data Creation Unit The new data creation unit NDTG creates new data as transmission data if there is no received packet used to create difference information. The case where there is no received packet used to create the difference information is as follows. That is, if NULL is entered in the “head address” field of the table CCT-1 in the cache management table CCT (see FIG. 15) and the received packet is not stored in the kernel buffer 200 (cache memory), the proxy application Either the APL has to create the transmission data completely new and the received packet is not available.

When there is no received packet used to create the difference information If there is no received packet used to create the difference information, the new data creation means NDTG uses the difference data structure DCB and the new data Create NDT. New data creation method for proxy application NDTG
Is the packet's memory address D in the difference data structure.
Set 1 to NULL. Also, since the new data is to be created, the "update area information D4" is set to NULL. The byte position D5 and bit position D6 of the data before update, the byte length D7 and bit length D8 of the data before update, and the byte length D9 and bit length D10 of the data after update are all set to NULL.

The address 11 for storing the new data is secured in the application buffer 300, and the new data is stored here as “data RRD after rewriting”. Address 11 is set to pointer D12 (D12 = address 1).
1). Since it is necessary to newly create the data portion and also newly create the IP / TCP header of the packet, "1" is set to "additional change presence / absence information D11 regarding the same packet". "New data flag D2" is set to 1. The “deletion data flag D3” is set to “0”. Thereafter, the difference information of the header part is created.

When there is a received packet used to create the difference information, but when creating new data. When the number of changed bits of the received data is large or when the number of changed bits is large, the new data creation determining unit NDCD Instructs creation of new data even if a received packet exists. As a result, the new data creation means NDTG sets the above-mentioned "when there is no received packet used for creating difference information".
According to the processing of the difference data structure DCB and new data NDT
Create In this case, the new data creation means NDTG sets the deletion flag D3 of the difference data structure DCB to 1 in order to delete the received packet in the kernel buffer KNL.
The following is a process of creating a difference data structure DCB and new data when creating new data although there is a received packet used to create difference information. First, the “memory address D1 of the packet” of the DCB is set as the address 2-1 and the packet PT1 to be deleted is specified. "Update area information D4" is set to NULL. The byte position D5 and bit position D6 of the data before update, the byte length D7 and bit length D8 of the data before update, and the byte length D9 and bit length D10 of the data after update are all set to NULL.

The address 11 for storing the new data is secured in the application buffer 300, and the new data is stored here as “data RRD after rewriting”.
Address 11 is set as pointer D12 (D12 = address 11). Since it is necessary to newly create the data portion and also newly create the IP / TCP header of the packet, "1" is set to "additional change presence / absence information D11 regarding the same packet". The “new data flag D2” is set to “1”. "Delete data flag D3" is set to 1 to discard the received packet stored in the cache memory (address 2-1).
To Next, a differential data structure DCB is similarly created for the packet PT2.

(15) New data creation judging means The new data creation judging means is NDCD. When the number of difference data structures created for one data (packet), that is, the number of areas to be updated is larger than the set number (See FIG. 7A) If the data length of the data to be updated for one piece of data is longer than the set length (see FIG. 7B), it is determined to create new data. Thereby, the difference information creating means DI
This has the effect of preventing the load of FC processing from increasing. According to the mechanism of (11) to (15) above, "difference information creation means"
Can be realized.

(C) Packet Transmission As described above, the proxy application APL performs predetermined processing on the received data, and then creates difference information DIF (difference data structure DCB, rewritten data RRD). . Hereinafter, a procedure of the terminal T3 from the creation of the difference information DIF to the transmission of the packet by the hardware HW will be described. FIG. 18 is a flow chart of a packet transmission process, and FIG. 19 is an explanatory diagram of creating transmission data. (16) Creation of transmission data The proxy application APL uses difference data creation means D
The IFC (FIG. 16) or the difference data structure DCB created by the new data creation means NDTG is stored at the address 10 (FIG. 19) in the application buffer 300, and the changed contents
(Rewritten data RRD) is stored at address 11.

(17) Data write message P If there is data to be transferred to the terminal T2, the proxy application APL determines whether a connection CN2 from the terminal T3 to the terminal T2 has not been established yet, Establish CN2 (see the description of (3) in the packet reception flow). Thereafter, the application APL sends the address 10 of the application buffer 300 as the “head address” to the transmission / reception-only interface of the connection CN2 (hereinafter referred to as socket 2).
Further, it sends out a "data write message P" with the "data length" of the rewritten data RRD on the address 11 as the "data length" as arguments.

(18) Copying of Difference Information to Kernel Upon receiving the “data write message P”, the kernel KNL copies the difference data structure DCB at address 10 to the address 20 in the kernel buffer 200 and With reference to the address indicated by the pointer D12 included in the data structure DCB, the rewritten data RRD of the address 11 is copied to the address 21. (19) Generation of update wait queue Next, the kernel KNL queues the update wait buffer RNB2.
Queue the address 20 in the (update queue).
The update wait queue is a queue for storing the head address of the differential data structure DCB, and an update unit UPD described later performs update processing sequentially from the head of this queue.

(20) Handling of Deleted Data Flag in Update Waiting Queue The kernel KNL checks whether the “deleted data flag D3” of the differential data structure DCB is 1 (20-1), and if it is 1 (deleted). For example, in the subsequent processing, the received packet on the address 2 is not accessed. Therefore, the difference data structure DCB
2-1 indicated by the "packet memory address D1"
Delete the data of (20-2). (21) Handling of New Data Flag in Update Waiting Queue Next, the kernel KNL checks whether “new data flag D2” of the differential data structure DCB is 1 (21-1).
If "new data flag D2" is 0, that is, "new data flag D2" is 0, and "deletion data flag D3"
Is 1, the difference data structure DCB is created in “when data is discarded”. Therefore, the address 20 is deleted from the update wait queue RNB2, and the address 20 is deleted.
The above difference data structure is also deleted and the packet transmission process ends (21-2)

On the other hand, if the “deletion data flag D3” is 0 in (20-1), the kernel KNL checks whether the “new data flag D2” of the difference data structure DCB is 1 (21-21). 3). In the check of (21-1) and (21-3), if the new data flag D2 = 1, the kernel KNL sends the address 21 indicated by the pointer D12 of the focused difference data structure DCB to the transmission socket buffer SSB2. And then
Data structure DCB on 20 and buffer RNB waiting for update
Delete the address 20 on 2. Thereafter, the kernel KNL performs protocol processing in the order of the addresses stored in the transmission socket buffer queue SSB2 and transmits new data (21-4).

(22) Operation of Updating Unit UPD In (21-3), if the new data flag D2 is 0, that is, the deleted data flag D3 = 0 and the new data flag D2 = 0
If so, transmission data is created by performing an update process using the difference information DIF and the received packets PT1 and PT2 stored in the cache memory. The update unit UPD has the address 2
Based on the differential data structure DCB stored in 0, data (packets PT1 and PT2) in the cache memory (addresses 2-1 and 2-2) are changed as follows, and necessary protocol processing is performed.

(23) Execution of Change Based on Differential Data Structure The updating unit UDC first refers to the “memory address D1 of the packet” of the differential data structure DCB and determines that the packet to be updated is the packet PT.
Recognizing that it is 1, the data reference pointer is moved to the area start position in the packet PT1 indicated by the “update area information D4”. For example, if the “update area information D4” indicates the TCP header area, the update unit UPP refers to the IP header length in the IP header (FIG. 35A) and adds the IP header length to the packet start position. It recognizes that the bit position is the start position of the TCP header. Next, the updating unit UPD moves the data reference pointer to the rear of the packet PTi by the number of bits obtained by multiplying the “data byte position D5 before update” by 8 and further adding the “same bit position D6”. In addition, "Data byte length D7 before update and same bit length D8" are replaced with "Data byte length after update.
D9 and the same bit length D10 ", the data length before and after the change is equal. Therefore, the updating unit UPD replaces the data of the designated data length portion after the bit position before the update of the packet PT1 with the “data RRD after rewriting”, and then (25)
Execute the remaining protocol processing of. If the data lengths are not equal, (24) “update processing involving data length change” is executed.

(24) Update Process with Data Length Change FIGS. 20 and 21 are explanatory diagrams of the update process when the packet length is changed. FIG. 20 shows the case where the packet length decreases and FIG. 21 shows the case where the packet increases. It is. One unit of the area divided in units of 1000 bytes is called a memory block. Cache memory is data used by applications
(Packet), and a header section 2b for storing information relating to the data section.
a is divided in units of 1000 bytes (in units of one memory block), and a header portion HD1 corresponding to each of the memory blocks MB1 to MB3.
~ HD3 is provided. Packets are stored in memory block units.

The headers HD1 to HD3 are written with a pointer indicating the data order, a data length in the corresponding memory block, a data length of the entire packet, a TCP / IP header, and pointer information to the corresponding memory block. It has become.

Processing when the packet length is reduced The entire data length is 2500 bytes, and the packet PT1 is divided into three memory blocks MB1 to MB3 like the packet parts PT1-1 to PT1-3 and stored. In such a case, when the data length after the update is shorter than the data length before the update, the update unit UPD performs the following update processing. However, it is assumed that 500 bytes of data are deleted from the 1200th byte of the data portion of the packet PT1. The update unit UPD checks whether “the data byte length D9 after the update + the same bit length D10” is smaller than “the data byte length D7 before the update + the same bit length D8”. If it is smaller, as shown in FIG. 20 (b), the changed part of the memory block MB2 where the data is to be rewritten is replaced with the new “data RRD after rewriting (NULL for deletion)” and the empty part (NULL
Part). Next, the data length of the header part HD2 is changed (1000 → 500). Further, the data length of the entire packet of the header part HD1 is changed to the changed data length (2500 → 2000).

Processing when the packet length increases The entire data length is 2500 bytes, and the packet PT1 is divided into three memory blocks MB1 to MB3 like the packet parts PT1-1 to PT1-3 and stored. In such a case, when the data length after the update is larger than the data length before the update, the update unit UPD performs the following update processing. However, the packet PT1
It is assumed that 500 bytes of data from the 1200th byte of the data portion is replaced with 1000 bytes of data. Update section
UPD is “Data byte length D9 after update + Same bit length D10”
Is larger than “data byte length D7 before update + same bit length D8”. If it is larger, it is determined whether a new memory block needs to be prepared for data rewriting. If it is necessary, a new memory block MB2 'and a header section HD2' are prepared as shown in FIG. Next, the data portion of the memory block MB2 is changed by "rewritten data", the changed data portion is written to the memory blocks MB2 and MB2 ', and the data lengths of these memory blocks are written to the header portions HD2 and HD2'. . Next, the “data length of the entire packet” in the header HD1 is changed to the changed data length (2500 → 3000). Also, the pointer values of the header sections HD2 and HD2 'are changed so that the data order is from memory block MB1 to MB2 to MB2' to MB3.

(25) Remaining Protocol Processing After the update processing of the data portion is completed, the transmission socket 2 uniquely determines the identification number of the TCP header for each packet.
The SEQ number of the header is determined, and the checksum of the TCP header and the checksum of the IP header are calculated by the arithmetic expression defined by the TCP / IP protocol, and are written into the header part. Also, a predetermined value is written in the remaining protocol fields shown in FIG. 35 (25-1). All TCP / IP
When the protocol processing is completed, address 2 (cache memory address) is added to the transmission socket buffer SSB2.
Is saved (25-2). If terminal T3 is in the hardware layer,
When a protocol processing unit such as a checksum calculation unit is provided, an operation result using the protocol processing unit can be used as a checksum of a TCP header or a checksum of an IP header.

(26) Judgment of Completion of Header Creation The address 21 or the address 2 is queued in the transmission socket buffer SSB2 according to the processing of the above (21-4) or (25-2). It is determined whether or not the TCP / IP protocol processing has been completed with reference to the TCP / IP header of the packet created in the storage area indicated by these addresses.

(27) Protocol processing of transmission socket The kernel refers to the transmission socket buffer SSB2, and if it is determined that the packet has already been subjected to the header processing, further processing of the packet on the socket is performed. Is not performed. However, when it is determined that the packet has not been subjected to the header processing, the socket pair is changed for the TCP / IP header, and the processing of (25-1) is performed to complete the protocol processing. After completing the protocol processing, the kernel KNL copies the contents of the transmission socket buffer SSB2 to the retransmission socket buffer SSB2 '. The retransmission socket buffer is a mechanism for realizing a retransmission mechanism that the TCP protocol has as a standard.

(28) Determination of MAC address After all the TCP / IP protocol headers have been created,
The kernel KNL determines the destination MAC header from the route selection table, and sets the MAC address of the terminal T3 to the source MA.
A MAC header having a C address is created and added to the packet (28-1). Next, the head address 2 of the memory storing the entire packet to which the MAC header is added is stored in the driver queue. After that, by the operation specific to the driver,
Finally, the packet addressed to the MAC address described above is copied to the hardware buffer 100 (28-2), and the packet is transmitted in the hardware layer (28-3). By the above (16) to (27), the "updating means UPD" can be realized.

FIG. 22 is a diagram showing the entire flow from reception of a packet to creation of difference information, and creation and transmission of a transmission packet. In the present embodiment, a system in which a proxy application instructs a change to a socket pair is illustrated. Generally, in a protocol relay process for relaying a connection, a socket pair is changed for all packets. For this reason, creating a differential data structure for all packets to be relayed increases the amount of data copied from the application buffer to the kernel buffer. Therefore, the application does not instruct the change of the socket pair, and the updating unit UPD can update the socket pair portion of the header by using the socket pair held as internal information in the socket 2. By doing so, the copy amount can be reduced.

In the above, an example in which the present invention is applied to TCP has been described. However, the present invention can be applied to other protocols such as UDP (User Datagram Protocol). However, since UDP does not have a retransmission buffer, an operation such as storing data in the retransmission buffer is unnecessary. The present invention can also be applied to a relay device (bridge BRG) that relays communication between terminals at layer 2 as shown in FIG. Since the bridge BRG does not need to terminate the connection, when relaying data by the application APL, the bridge BRG determines whether or not to permit passing data, and performs only one of data passing and discarding. For this reason, the difference information creation unit DIFC always creates one of the difference information DIF described in (13) “when data is not changed” and “when data is discarded”, and the new data creation unit NDTG Since no new data is created, the new data creation unit NDTG is not required. Therefore, since it does not include the difference information, the rewritten data RRD, and the new data, the amount of data copied from the application APL to the lower layer can be significantly reduced. As described above, the amount of data to be copied from the application buffer to the kernel buffer can be reduced, and the redundancy of data copy can be significantly reduced.

(B) Second Embodiment In the first embodiment, the cache management table CCT for specifying the correspondence between the data in the application buffer and the packet group in the kernel buffer (cache memory) is created in the application buffer 300. The table can also be created in the kernel buffer 200.
The second embodiment is an embodiment in which the cache management table CCT is provided in the kernel buffer 200, and realizes the processing described in FIG.

FIG. 24 is an explanatory diagram of the entire flow from reception to transmission of a packet according to the second embodiment, and FIG. 25 is a configuration diagram of a differential data structure DCB according to the second embodiment. In the second embodiment, the proxy application APL has a differential data structure D
When creating the CB, the data change location is specified by the address of the application buffer 300. The updating unit UPD of the kernel to which the difference information DIF has been transmitted converts the address of the application buffer into the address of the cache memory with reference to the cache management table CCT, whereby the received packet having the changed portion and the change in the packet are changed. Acquires the position and changes the received packet to the difference information DI
Create the transmission data by replacing the contents of F with the changes. Hereinafter, processing different from the first embodiment will be described.

Changes in packet reception flow Steps in packet reception flow of the first embodiment
In the second embodiment, the following steps (29) and (30) are performed instead of the steps (6) to (10) (FIG. 13). (29) Data read message D2 When the proxy application APL associated with the socket 1 in advance receives the message A from the socket 1 in step (5), it prepares a memory area for reading data at address 4 in the application buffer 300. Then, a "data read message D2" having "data length" and "address 4" as arguments is sent to the kernel KNL.
Send to

(30) Writing to Application Buffer by Kernel Kernel that Received “Data Read Message D2”
The KNL writes data on the cache memory (address 2) indicated by the address stored at the head of the reception socket buffer RSB1 as one piece of data at the address 4 of the application buffer 300 excluding the protocol header. At the same time, the kernel KNL creates a table CCT-h (see FIGS. 14 and 15) in the kernel buffer 200, and notifies the “head address” field and the “data length” field of the message C2 from the application using the message D2. Write address 4 and data length. Then the table
Write the address 2 and the data length to be read in the “head address” field and “data length” field of CCT-1.

When there are a plurality of packets to be written,
Each time the i-th packet PTi is written to the address 4, a predetermined address and data length are written in the above fields of the table CCT-i, and the number of packets i is written in the "number of packets" field of the table CCT-h. Do. After writing all data, NULL is written in the “next table pointer” field of the table CCT-N, and the creation of the cache management table CCT is completed. Thereafter, the read address is deleted from the receive socket buffer RSB1, but the received packet PTi on address 2 is deleted.
Is not deleted, the copied data is cached at address 2. That is, the address 2 of the kernel buffer 200 becomes a cache memory area.

Changes in Packet Transmission Flow In the second embodiment, unlike the first embodiment, the application is not aware of the packet in the kernel buffer. For this reason, the application executes the cache management table in steps (11) to (15) of creating the difference data structure DCB.
Instead of referring to the CCT, the data change location is specified by the address of the application buffer that stores the data. The difference data structure DCB at this time is, as shown in FIG.
The configuration is such that the “packet memory address D1” is replaced with the “data memory address D1 ′”. That is, for example, if data is stored at address 4, D1 '
= Address 4. -Changes in the packet transmission flow In the packet transmission process, the steps of the first embodiment
By adding the process of the following step (31) before (23),
The data difference information specified by the difference data structure DCB is converted into packet difference information.

(31) Specification of Packet Group Corresponding to Data in Application Buffer When the updating unit UPD of the kernel KNL receives the differential data structure DCB from the application, the “data memory address D1 ′” of the differential data structure DCB is received. It is necessary to specify the memory address of the cache memory (address 2) in which the packet group corresponding to the data on the address 4 indicated by is stored. Therefore, the updating unit UPD refers to the cache management table CCT in the kernel buffer 200, searches the table for a table CCT-h including the address 4 in the “head address” field, and includes the table CCT-h in the table. The "start address" field of the tables CCT-1, CCT-2, .., CCT-N is extracted. Thereby, a packet group corresponding to the data of the address 4 can be specified.

Further, the packet including the changed portion indicated by “byte position D7 and bit position D8 of data before update” and “byte length D9 and bit length D10 of data before update” in differential data structure DCB is stored in table CCT- 1, CCT-2, .... Specify by referring to the data length in the "data length" field of CCT-N. For example, when the table CCT-2 includes the changed portion, the “data memory address D of the data
"1 '" is changed to "Start address value" of table CCT-2. In addition, the message "Byte position D7 and bit position D
Rewrite "8" to the position specification based on the packet to be changed in the cache memory. In this way, in the subsequent processing, D1 'can be treated as a "packet memory address", and the packet can be updated and transmitted by the same procedure as in the first embodiment.

When the changed portion specified by one difference data structure DCB extends over a plurality of packets, a plurality of difference data structures are created by copying or the like, stored in the update waiting buffer RNB2, and the “update process” is executed. , The update is performed on all the packets to be updated in the cache memory. If the difference data structure relates to a change in a protocol header such as a change in a socket pair, the difference data structure is copied for all packets. By having the above processing procedure, there is an advantage that the proxy application has the same function as that of the first embodiment, and the proxy application can create the difference information without being conscious of the packet, thereby facilitating the creation of the program. In the above-described first and second embodiments, the hardware buffer and the kernel buffer are separately provided. However, the hardware buffer can be directly accessed from the kernel KNL, that is, the hardware buffer and the kernel buffer are provided. And the first and second embodiments can be modified as shown in FIGS.

(C) Third Embodiment The third embodiment is applicable to the case where the relay device and the application server are different nodes, and realizes the application processing described with reference to FIG. FIG. 26 is an explanatory diagram of the third embodiment. The application (relay application) APL1 on the relay device RT that has received the connection from the connection CN1 is an application server ASV
A connection CN10 is established between the application server ASV and the data is transferred to the proxy application APL2 operating on the application server ASV, and after application processing, transmission data is created based on the difference information returned from the connection CN10 and transmitted to the connection CN2. In this way, data is relayed.

Hereinafter, differences from the second embodiment will be described.
First, after finishing the packet receiving process of step (30), the relay application APL1 performs the following processes of steps (32) and (33). (32) Transfer means to the proxy application According to the reception processing of the packet from the connection CN1,
Upon receiving the data, the relay application APL1 establishes a connection CN10 with the application server ASV by the connection opening means, and transfers the data on the address 4 of the application buffer to the application server ASV via the connection CN10. At this time, it is assumed that the data is stored at the address 44 on the application server ASV.

Further, as an additional process, the relay application APL1 creates a conversion table CVTB for storing that the data at the address 4 has been transferred via the connection CN10. The conversion table CVTB is a table that stores the correspondence between the “head address” of the application buffer that stores data in the relay node RT and the identifier (file identifier) assigned to the socket of the connection CN10 that transmits the data. (See FIG. 10). Here, “address 4” and “fd0” are stored. The socket identifier means a numerical value capable of uniquely identifying a socket. For example, when the application APL1 creates a socket, a different numerical value is used for each socket obtained from the kernel.

(33) Means of Transfer from Proxy Application The proxy application APL2 on the application server ASV receives the data received in the step (32).
(Data on the address 44) is subjected to predetermined data processing, and the difference information creating unit DIFC creates difference information DIF (difference data structure DCB and change contents RRD) based on the data processing result. The difference information creation unit DIFC creates the difference data structure DCB and the rewritten data RRD as in the second embodiment. This difference data structure DCB has difference information with respect to the data on the address 44, and “data memory address D
"1 '" is an address 44. Next, the proxy application APL2 transmits the created difference information DIF from the connection CN10 to the relay application APL1 on the relay device RT. Here, the difference data structure DCB is stored at the address 64 of the application buffer on the relay device.

(34) Means for Modifying the Differential Data Structure The “data memory address D1 ′” of the differential data structure DCB stored at the address 64 is the application server.
The address 44 on the ASV is shown. Therefore, it is necessary to convert the address 44 into the data memory address (address 4) of the relay device RT. Therefore, the difference information correcting unit DICR of the relay application APL1 acquires the address 4 corresponding to the file identifier fd0 by referring to the conversion table CVTB, and uses the address 4 to address the “data memory address D1 ′” of the difference data structure DCB. The address is changed from 44 to address 4. Thereafter, the relay application APL1 transmits the corrected difference data structure DCB to the kernel KNL, and the updating unit UPD of the kernel KNL stores the difference information DIF, the cache management table CCT, and the cache memory (address 2). The transmission data is created using the stored reception packet.
Lower layer of relay equipment (kernel KNL, hardware H
The processing in W) is the same as the processing in step (31) of the second embodiment.

First Modification In the third embodiment described above, the address of the application server ASV is converted to the address of the relay device RT using the conversion table CVTB. However, the third embodiment may be configured using the identification table RGTB. FIG. 27 is a configuration diagram of such a conversion example, which implements the processing of FIG. The purpose of the identification table RGTB is the same as that of the conversion table CVTB. The relay application APL1, which stores the received data at the address 4, transfers the data to the application server ASV via the connection CN10, assigns a unique identification number to the received data at the address 4, and assigns the identification number and the address 4 Are stored in the identification table RGTB. The proxy application APL2 uses the difference information creation means DIFC
As a result, as shown in FIG. 28, a difference data structure DCB including the identification number D13 is created as a new member, and transmitted to the relay application APL1. Relay application APL1
When the difference information is transferred, the difference information correcting unit DICR refers to the identification number D13 included in the difference data structure DB to obtain the address 4 from the identification table RGTB, and uses the address 4 to obtain the difference data structure DCB. "Data memory address D1 '" is changed from address 44 to address 4. Thereafter, the relay application APL1 sends the modified difference data structure DCB to the kernel KNL, and the updating unit UPD of the kernel KNL updates the difference information DIF and the cache management table CCT.
The transmission data is created using the received packet stored in the cache memory (address 2).

Second Modified Example Although the third embodiment has been described based on the second embodiment, the third embodiment may be described based on the first embodiment. FIG. 29 is a configuration diagram of such a modification. In the first embodiment, the received data and the cache management table CCT are stored in the application buffer of the relay device RT. Therefore, instead of creating the conversion table CVTB, the relay application APL1 stores the cache management table CCT in the application server ASV.
Separately. The cache management table CCT transferred from the relay application APL1 indicates the correspondence between the data in the application buffer (address 4) on the relay device RT and the packets stored in the cache memory (address 2). Therefore, the table correction unit TBCR corrects the data into the table CCT 'so as to indicate the correspondence between the data in the application buffer (address 44) on the application server ASV and the packet on the cache memory.

Next, the proxy application APL2 performs processing such as security check on the received data.
After the check processing, the application APL2 creates difference information using the difference information creation unit DIFC. That is, the difference information creation unit DIFC stores a packet corresponding to the data portion that needs to be changed based on the check result in the cache management table CC.
It specifies with reference to T 'and creates the change.
The application APL2 transmits the difference information DIF to the application APL1 of the relay device RT via the communication unit 10. The application APL1, which has received the difference information DIF,
Difference information in the kernel buffer accessible by the update unit UPD
DIF copy. Upon receiving the difference information DIF from the application APL1, the updating means UPD specifies the packet PTi on the cache memory to be changed by the difference information DIF, and then specifies the changed portion in the packet PTi, and Is replaced with the changed content, and this is used as transmission data.

Third Modification In the relay system in which the relay device (node 1) and the application server (node 2) constitute different nodes,
Any of the following modifications: 1. providing a cache memory in the application layer of node 1; 2. providing an update unit in the application layer on node 1; Sharing of the buffer of the lower layer (kernel, hardware) with the application layer of the node 1 can be adopted. If the buffers are shared, copying between the buffers can be eliminated.

As described above, according to the third embodiment and its modifications,
Even if the relay device and the proxy application are installed on different nodes, it is possible to greatly reduce the redundancy of the copy processing within the node and mainly between the nodes. As described above, the present invention has been described with reference to the embodiments. However, the present invention can be variously modified in accordance with the gist of the present invention described in the claims, and the present invention does not exclude these.

[0142]

As described above, according to the present invention, it is possible to perform application processing such as security check on received data located at the boundary of a network, and transmit the data to the next terminal on a communication path. In the relay device that relays data, high-speed communication can be performed while eliminating redundancy of copy processing as much as possible.
Further, according to the present invention, when it is determined not to transmit data to the outside, it is only necessary to instruct the lower layer to delete from the upper layer, and the processing time can be reduced. Further, when no change is made to the received data, the processing time can be shortened only by instructing no change from the upper layer to the lower layer. When the number of changed parts for the same packet is larger than the set number,
Alternatively, when the total sum of the changed data lengths for the same packet is larger than the set number, new transmission data is created, so that the processing time can be reduced. Furthermore, according to the present invention, even when an application that performs data processing such as a security check is provided in an application node different from the relay node, the redundancy of copy processing between nodes is greatly reduced. High-speed communication can be performed.

[Brief description of the drawings]

FIG. 1 is a schematic explanatory diagram of a first application process of the present invention.

FIG. 2 is a schematic explanatory diagram of a second application process of the present invention.

FIG. 3 is a schematic explanatory diagram of a third application process of the present invention.

FIG. 4 is a schematic explanatory diagram of a fourth application process of the present invention.

FIG. 5 is a schematic explanatory diagram of a fifth application process of the present invention.

FIG. 6 is a schematic explanatory diagram of a sixth application process of the present invention.

FIG. 7 is an explanatory diagram of switching determination between creation of difference information and creation of new data.

FIG. 8 is a schematic explanatory diagram of an eighth application process of the present invention.

FIG. 9 is a schematic explanatory diagram of a ninth application process of the present invention.

FIG. 10 is a diagram illustrating the operation of a conversion table.

FIG. 11 is a schematic explanatory diagram of a tenth application process of the present invention.

FIG. 12 is a conceptual diagram of the relay of the present invention.

FIG. 13 is a reception flow of a packet.

FIG. 14 is an explanatory diagram showing a correspondence relationship between data in a kernel buffer and data in an application buffer.

FIG. 15 is a cache management table.

FIG. 16 is an explanatory diagram of a difference information creation process.

FIG. 17 is an explanatory diagram of a difference data structure.

FIG. 18 is a packet transmission flow.

FIG. 19 is an explanatory diagram of transmission data creation.

FIG. 20 is an explanatory diagram of an update process when a packet length is changed (decreased).

FIG. 21 is an explanatory diagram of an update process when a packet length is changed (increased).

FIG. 22 is an explanatory diagram of an entire flow from reception to transmission of a packet.

FIG. 23 is a conceptual diagram of a bridge relay.

FIG. 24 is an explanatory diagram of the overall flow from packet reception to transmission in the second embodiment.

FIG. 25 is an explanatory diagram of a differential data structure according to the second embodiment.

FIG. 26 is an explanatory diagram of a relay process according to the third embodiment.

FIG. 27 is a first modification of the third embodiment.

FIG. 28 is an explanatory diagram of a difference data structure with an identification code.

FIG. 29 is a second modification of the third embodiment.

FIG. 30 is an explanatory diagram of a conventional network environment.

FIG. 31 is an explanatory diagram of another conventional network environment.

FIG. 32 is an explanatory diagram of a hierarchical configuration of an Internet protocol.

FIG. 33 is a diagram illustrating a message configuration of each layer.

FIG. 34 is a conceptual diagram of relaying.

FIG. 35 shows a TCP / IP protocol header.

FIG. 36 is a conventional packet reception flow.

FIG. 37 is a conventional packet transmission flow.

FIG. 38 is an explanatory diagram of creation and transmission of transmission data.

[Explanation of symbols]

 100 Hardware buffer 200 Kernel buffer 300 Application buffer CSH Cache means CCTC Cache management table creation means UPD Update means CCT Cache management table DIFC Difference information creation Means DIF

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5K030 GA02 HA08 HB28 HD01 KA02 KX11 LD01 LE11 5K033 AA02 BA13 CB08 CC01 DA05 DB12 DB18 5K034 AA02 BB06 CC01 DD03 EE11 FF11 HH01 HH14 HH17 HH18 9A001 BZ03 CC03 LL03

Claims (17)

[Claims] 1. A relay device for performing predetermined data processing on data received from the outside via a lower layer in an application layer as an upper layer, and transmitting data according to the processing result to the outside via the lower layer. A cache memory of a lower layer for storing data received from the outside, a cache management table indicating a correspondence relationship between data copied to a buffer of an application layer and data stored in the cache memory, and stored in the buffer. Means for creating a table, after data processing, specifies data in the cache memory that needs to be changed using the correspondence held in the cache management table, and has the data specifying information, the changed location of the data, and the changed content. Difference information in the application layer that creates difference information A relay unit that updates data stored in a cache memory using the difference information, and uses a result of the update as transmission data in a lower layer. 2. The relay device according to claim 1, wherein said cache memory is provided in a buffer of a hardware layer accessible to an updating unit. 3. The relay device according to claim 1, wherein the cache memory stores the received data in packet units, and the table creating means stores the copy destination data in the application buffer and the packet in the cache memory. A cache management table indicating the correspondence is created, and the difference information creating unit identifies the packet in the cache memory that needs to be changed using the association held in the cache management table, and determines the packet identification information and the packet identification information. The update unit generates difference information having the changed portion and the changed content of the above, and the updating unit updates the packet stored in the cache memory using the difference information, and uses the update result as transmission data. 4. A relay device for performing predetermined data processing on data received from the outside via a lower layer in an application layer, and transmitting data according to the processing result to the outside via a lower layer. A cache memory of a lower layer for storing received data, a table for creating a cache management table indicating a correspondence between data copied to a buffer of an application layer and data stored in the cache memory, and storing the table in a buffer of a lower layer Creating means, application data difference information creating means for identifying data that needs to be changed among the data copied to the application buffer, and creating difference information including the data specifying information, the changed location of the data, and the changed content; And kept in the cache management table And uses the relationship to identify the data requiring changes stored in the cache memory, a relay apparatus characterized by having, updating means in the lower layer to transmit data and update the data. 5. The relay device according to claim 4, wherein the cache memory is provided in a buffer of a hardware layer accessible to an updating unit. 6. The relay device according to claim 4, wherein the cache memory stores the data received in packet units, and the table creating means stores the copy destination data in the application buffer and the packet in the cache memory. The update unit creates a cache management table indicating the correspondence, identifies the packet that needs to be changed using the difference information and the correspondence held in the cache management table, updates the packet, and updates the packet with the transmission data. It is characterized by the following. 7. The relay device according to claim 1, wherein the difference information creating unit includes a deletion instruction in the difference information when data in the cache memory is not transmitted to the outside, and the updating unit includes the updating unit. Data in the cache memory is deleted by a deletion instruction. 8. The relay device according to claim 1, wherein the difference information creating unit causes the difference information to include a no-change instruction when the data in the cache memory is not changed, and the updating unit includes: According to the non-change instruction, data in the cache memory is directly used as transmission data. 9. The relay device according to claim 1, wherein the difference information creating means includes a new instruction in the difference information when newly creating transmission data, and the updating means makes the difference information in accordance with the new instruction. Data included in the information is newly transmitted data. 10. The relay device according to claim 3, wherein, when the number of changed portions for the same packet is larger than the set number, the application creates new transmission data, and the difference information creating means stores the difference information in the difference information. A new instruction and the transmission data are included, and the updating means uses the transmission data included in the difference information according to the new instruction as new transmission data. 11. The relay device according to claim 3, wherein when a total sum of changed data lengths for the same packet is longer than a set length, the application creates new transmission data, and the difference information creating means includes the difference information creating unit. And a transmission instruction included in the difference information according to the new instruction. 12. The first node transfers externally received data to a second application of a second node by a first application, and the second node performs predetermined data processing on the received data by the second application. Transferring the processing result to the first application of the first node,
A relay device for generating a transmission data based on the transferred processing result and transmitting the data to the outside, wherein the first node includes: a lower-layer cache memory for storing data received from the outside; and a buffer for an application layer. Table creation means for creating a cache management table indicating the correspondence between the data copied to the cache memory and the data stored in the cache memory; the received data and the cache management table between the first and second applications; Communication means for transmitting and receiving difference information and the like; updating means for updating data stored in the cache memory using the difference information transferred from the second application, and updating the lower layer with the update result as transmission data. , The second node receives data sent from the first node, A buffer of an application layer for storing the cache management table, and a table correction means for correcting the correspondence of the cache management table to the correspondence between the data copied to the application buffer of the second node and the data stored in the cache memory. After the data processing by the second application, the data in the cache memory that needs to be changed is specified by using the correspondence relationship held in the corrected cache management table, and the data specifying information, the changed location of the data, and the change A relay device, comprising: upper layer difference information creating means for creating difference information having contents; received data and a cache management table between the first and second applications; and communication means for sending and receiving difference information. . 13. The relay device according to claim 12, wherein the cache memory stores the data received in packet units, and the table creating means stores the copy destination data in the application buffer of the first node and the packet in the cache memory. A cache management table indicating the correspondence is created, and the table correction unit corrects the correspondence in the cache management table to a correspondence between the copy destination data in the application buffer in the second node and the packet in the cache memory, The information creating unit identifies a packet in the cache memory that needs to be changed using the correspondence relationship held in the corrected cache management table, and compares the packet identification information and the difference information having the changed portion and the changed content of the packet. The update unit creates a cache using the difference information. Update the packet stored in Yumemori, the update result and transmits data, and wherein the. 14. The first node transfers externally received data to a second application of a second node by a first application, and the second node performs predetermined data processing on the received data by the second application. Transferring the processing result to the first application of the first node,
A relay device for generating a transmission data based on the transferred processing result and transmitting the data to the outside, wherein the first node includes: a lower-layer cache memory for storing data received from the outside; and a buffer for an application layer. Table creation means for creating a cache management table indicating the correspondence between the data copied to the (first application buffer) and the data stored in the cache memory, between the first and second applications via a predetermined connection Communication means for transmitting and receiving data and difference information as the processing result; a conversion table indicating a correspondence relationship between a head address of the data copied to the first application buffer and the identifier of the connection; Difference information transmitted Using the data head address of the second application buffer included in the difference information and the data head address of the first application buffer obtained from the conversion table. An update unit in a lower layer that specifies data that needs to be changed and that is stored in the cache memory using the correspondence relationship held in the cache management table and updates the data to be transmission data. The two nodes specify a buffer (second application buffer) for storing data sent from the first node, data in the second application buffer that needs to be changed after data processing by the second application,
Upper layer difference information creating means for creating difference information having the data specifying information, the changed portion of the data, and the changed content, and transmitting and receiving data and difference information between the first and second applications via the connection A relay device, comprising: communication means. 15. The relay device according to claim 14, wherein the cache memory stores the data received in packet units, and the table creating means stores the copy destination data in the application buffer of the first node and the packet in the cache memory. A cache management table indicating the correspondence is created. The difference information creation unit identifies data in the second application buffer that needs to be changed, and creates difference information including the data identification information, the changed location in the data, and the changed content. The difference information correcting means corrects the difference information by using the data start address of the second application buffer included in the difference information and the data start address of the first application buffer obtained from the conversion table, and the updating unit corrects the difference information. Difference information and correspondence held in the cache management table Identify the packet requiring changes by using the engagement, and transmits data to update the packet, characterized in that. 16. The first node transfers externally received data to a second application of a second node by a first application, and the second node performs predetermined data processing on the received data by the second application. Transferring the processing result to the first application of the first node,
A relay device for generating a transmission data based on the transferred processing result and transmitting the data to the outside, wherein the first node includes: a lower-layer cache memory for storing data received from the outside; and a buffer for an application layer. Table creation means for creating a cache management table indicating the correspondence between the data copied to the (first application buffer) and the data stored in the cache memory; data to which an identifier is assigned between the first and second applications Communication means for transmitting and receiving difference information as a result of the processing, an identification table indicating a correspondence between a head address of data copied to the first application buffer and an identifier of the data, and a connection from the second application via the connection. Transfer the difference information to the difference information Difference information correcting means for correcting using the data head address of the second application buffer to be stored and the data head address of the first application buffer obtained from the identification table; and holding the corrected difference information and the cache management table. Updating means in a lower layer for identifying data that needs to be changed and stored in the cache memory using the corresponding relationship, and updating the data to be transmission data. A buffer for storing data sent from one node (second application buffer); after processing data by the second application, specify data in the second application buffer that needs to be changed;
Upper layer difference information creating means for creating difference information having the data specifying information, the changed portion of the data, and the changed content, and transmitting and receiving data and difference information between the first and second applications via the connection A relay device, comprising: communication means. 17. The relay device according to claim 16, wherein the cache memory stores the received data in packet units, and the table creating means stores the copy destination data in the application buffer of the first node and the packet in the cache memory. A cache management table indicating a correspondence relationship is created. The difference information creating unit identifies data in the second application buffer that needs to be changed, and determines a difference including the data identification information, the changed location in the data, the changed content, and the identifier. Information is created, and the difference information correcting means corrects the difference information using the data start address of the second application buffer included in the difference information and the data start address of the first application buffer obtained from the identification table. Modified difference information and stored in the cache management table Identify the packet requiring changes by using the in which the corresponding relationship, and transmits data to update the packet, characterized in that.