KR20080060054A - Distributed flow control system - Google Patents

Abstract

A distributed flow control system is provided to enable a system to solve traffic jam and realize efficient data communication by distributing load concentrated on a server without any load distribution device like L(Level)4 and L7 switches. A network receiver(610) receives a packet including a first sequence map, which is generated from sequence numbers received from a transmission server by a client. A controller(620) calculates network delay between the transmission server and the client by determining whether the sequence numbers included in the first sequence map is omitted. A network transmitter(630) transmits a flow control packet, which includes a command for controlling packet transmission from the transmission server to the client according to the calculated network delay between the transmission server and the client, to the transmission server. The controller calculates the network delay, and determines network transmission error between the transmission server and the client by storing the last sequence map and comparing the first sequence map with the last sequence map.

Description

Distributed Flow Control System {DISTRIBUTED FLOW CONTROL SYSTEM}

1 is a diagram illustrating a configuration of a service processing system according to an example of the present invention.

2A illustrates a service request packet according to the present invention.

2B is a diagram illustrating first information according to the present invention.

2C is a diagram illustrating a response packet according to the present invention.

3 is a diagram illustrating a configuration of a distributed flow control system according to another example of the present invention.

4A is a diagram illustrating a first sequence map according to the present invention.

4B is a diagram illustrating a second sequence map according to the present invention.

5A is a diagram illustrating an example of a first sequence map according to the present invention.

5B is a diagram illustrating an example of a second sequence map according to the present invention.

5C illustrates another example of the second sequence map according to the present invention.

5D illustrates another example of the second sequence map according to the present invention.

6 is a diagram illustrating a receiving server according to an example of the present invention.

7 is a flowchart illustrating a distributed flow control system according to an example of the present invention.

<Description of the symbols for the main parts of the drawings>

100: transmit and receive network

110: client

120: receiving server

130: sending server

140: service request packet

150: first information

160: response packet

The present invention relates to a service processing system, and more particularly, to flow control in the case of efficiently handling a service request from a client by distributing a load concentrated on a server in a network.

An environment where multiple computers share work and is called a distributed processing environment. In a distributed processing environment, resources can be shared or exchanged with each other over a network. In a network composed of a client and a server, a user's terminal is a client and a host computer is a server. That is, it means a structure in which data requesting is possible between a client requesting a service and a server providing a service corresponding thereto. The client connects to the server for tasks that are intended to concentrate data or share resources. However, the network environment based on the Internet is becoming more and more complicated. For example, in order to receive a service called HTTP (hypertext transfer protocol), user clients connect to a real HTTP server on a space in the Internet and receive a desired service. In other words, if the server is only connected to the network environment of the Internet, users can receive the service without regard to its physical location or address. However, from the server's point of view, the problem is more complicated than that of the user clients. In order to provide services reliably and continuously to many users, the server must have a large ability to process them. Instead of upgrading to more powerful servers to provide more accurate service to more users, switches are installed that install multiple servers and assign service requests from clients to the multiple servers. However, as the traffic that has to be handled in the network has soared, congestion has begun to occur frequently in switches that must handle the headers of all packets. If congestion occurs at the switch, the switch cannot process the received packet, which in turn causes a retransmission of the client, making the situation even worse.

As a solution, a connection distribution scheme was used. The connection distribution method is a server load balancing method using a Layer-4 (L4) switch, and analyzes the load of each connected server and distributes the load to each server. The server load balancing method by the L4 switch does not rely on the domain name server (DNS) and binds multiple servers to one virtual IP (VIP) address to inform users of only VIP and not only IP address but also four layers. In the TCP / UDP port, even if a specific server fails, correct load balancing among the servers is possible.

However, in the connection distribution scheme using the conventional L4 switch, the L4 switch receives all packets and extracts a connection tuple of the corresponding packet so that a mapping table exists in the connection tuple. In this case, the server that matches the tuple is extracted from the mapping table and the packet is delivered to the server. Therefore, since all packets must pass through the L4 switch, the bottleneck occurs because all loads are concentrated on the L4 switch. In addition, since not only the user request packet received from the client but also the response packets generated by the plurality of servers and delivered to the client pass through the L4 switch, a bottleneck occurs in the L4 switch.

Therefore, the present invention allows the receiving server to respond to a service request received from the client and to be dedicated to the sending server in response to the client even without the use of devices such as L4 and L7 switches, and the receiving server is not involved in subsequent data transmission. By distributing the load, the bottleneck is eliminated and efficient data communication is possible, and the flow control method in such a distributed system is proposed.

In addition, in the transmission and reception of data between the client and the reception server or the transmission server, the client and the reception are controlled by using a flow control that adjusts the data flow so that the transmission speed of the transmission side is not slowed down due to the speed delay of the reception side. A distributed flow control system capable of efficient data communication between a server and a sending server is proposed.

In addition, in the transmission and reception of data between the client and the receiving server or the sending server, an error control for retransmitting unreceived data to the client side enables distributed data communication between the client, the receiving server, and the sending server. We would like to propose a flow control system.

The present invention has been made to improve the prior art as described above, distributed flow control system that can solve the bottleneck and efficient data communication by distributing the load concentrated on the server without the use of load balancers such as L4, L7 switch The purpose is to provide.

In addition, the present invention is intended to provide a distributed system in which the reception server is dedicated to receiving data from the client, and the transmission server is dedicated to transmitting data to the client, so that no bottleneck occurs at a specific point. In particular, the conventional L4 switch or the like, as well as the packet to be transmitted all received through the L4 switch, the present invention, the received packet is received only to the receiving server, the transmitted packet is not sent through the receiving server Therefore, an object of the present invention is to provide a distributed flow control system in which a problem of a bottleneck according to the prior art is solved because it is transmitted by a transmission server.

In addition, in the present invention, since the network between the client and the receiving server and the network between the client and the sending server are separated from each other, the delay state of the network between the client and the receiving server is determined through comparison between sequence maps composed of sequence numbers, Determining the network delay between servers through the sequence number missing from the sequence map, adjusting the transmission and reception speeds of the network between the client and the receiving server and the network between the client and the sending server, respectively, and retransmitting packets received to the client. It is possible to provide a distributed flow control system capable of transmitting and receiving data efficiently and stably according to the delay state of each network.

In order to achieve the above object and solve the problems of the prior art, a distributed flow control system according to an aspect of the present invention, the transmission server and the sequence for generating a first packet and the sequence number associated with the first packet to the client And a receiving server that receives from the client a second packet comprising a first sequence map generated by the client using a number. The first sequence map is generated by grouping the sequence number by a predetermined number. The receiving server sends a flow control packet to the transmitting server, the control server comprising a command to determine whether a sequence number included in the first sequence map is missing and to control packet transmission from the transmitting server to the client. Send. The receiving server stores the most recently received final sequence map and compares the first sequence map and the final sequence map received from the client to adjust packet transmission from the client to the receiving server. Send a flow control packet containing the command to the sending server.

According to still another aspect of the present invention, in response to a service request received from the client to the transmission server, the transmission server generates a reply packet for the service request packet and transmits the response packet to the client. The destination network address of the response packet is the source network address of the service request packet, the source port number of the response packet is the destination port number of the service request packet, and the destination port number of the response packet is of the service request packet. The source port number.

According to still another aspect of the present invention, a network receiver for receiving a packet from the client, the packet including a first sequence map generated by a client using a sequence number received from a transmission server, and included in the first sequence map. And a controller configured to determine whether a sequence number is missing and calculate a network delay between the transmitting server and the client, and from the transmitting server to the client according to the calculated network delay between the transmitting server and the client. And a network transmitter for transmitting a flow control packet to the transmission server, the command including a command to regulate packet transmission of the packet.

According to another aspect of the present invention, the control unit stores the most recently received final sequence map, and compares the first sequence map and the last sequence map received from the client, the client and the receiving server Calculate the network delay between. The apparatus may further include a network transmitter configured to transmit a flow control packet including a command to control packet transmission from the client to the receiving server according to the determination result of the controller.

According to another aspect of the invention, the first sequence map is a sequence number for the nkth service packet (n and k are integers greater than 1) received from the transmitting server from the sequence number for the nth service packet It includes. The network receiver may generate a second sequence map generated by the client including a sequence number for an nxk-th service packet (where x is an integer of 1 or more) to a sequence number for an nx-th service packet. Receive. The controller compares the first sequence map and the second sequence map to determine a change in network delay between the client and the transmission server.

According to still another aspect of the present invention, a packet having a destination network address as a representative network address assigned to the transmission server and the reception server is received in the network reception unit, and the first packet received by the client from the transmission server. The source network address of is the representative network address. The network receiver receives a service request packet from the client, and the network transmitter transmits a source network address, a source port number, and a destination port number of the service request packet to the transmission server. The sending server is physically separated from the receiving server and connected to the network.

According to another aspect of the present invention, the first sequence map using the sequence number for the n-th service packet from the sequence number for the nk-th service packet (the n and k is an integer of 1 or more) received from the transmitting server Generating and transmitting the first sequence map to a receiving server that is physically separate from the transmitting server. A second sequence map is generated by using a sequence number for an nxkth service packet (where x is an integer equal to or greater than 1) received from the transmitting server to a sequence number for an nxth service packet. And transmitting to the receiving server together with a first sequence map, the computer readable recording medium having recorded thereon a program for executing the method.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings and the contents described in the accompanying drawings, but the present invention is not limited or limited to the embodiments. Like reference numerals in the drawings denote like elements.

1 is a diagram illustrating a configuration of a service processing system according to an example of the present invention.

As shown in FIG. 1, the service processing system according to the present invention includes a receiving server 120 and a transmitting server 130. The reception server 120 and the transmission server 130 constitute a transmission / reception network 100 and are connected to the client 110 through a network.

The reception server 120 is connected to the network with the client 110 to receive a service request packet 140 from the client 110. The service request packet 140 refers to a packet sent by the client 110 to request a service to a specific server in order to use data information of the server. The service request that the client 110 may make to the server side may include a connection to a website, a download of data, a streaming service of data, and the like. For example, when the client 110 requests a download of data to a specific server, the client 110 may receive a desired service by transmitting a service request packet to the specific server and downloading data from the specific server.

The service request packet 140 includes an IP address and port of a client side making a service request, and an IP address and port of a server side receiving a service request. Since the reception server 120 is set to the representative IP address of the transmission / reception network 100, the reception server 120 is a representative of the transmission / reception network 100 among the service request packets 140 transmitted by the client 110. Receive the service request packet 140 received at the IP address.

In particular, when the IP address of the receiving server 120 is set as the representative IP address of the transmission / reception network 100, all the service request packets 140 from the client 110 having the representative IP address as the destination address are all received server 120. Is received). Receiving server 120 is physically separated from the transmitting server 130 in the transmission and reception network 100 is responsible for receiving the service request packet 140 from the client 110, the service request packet 140 Does not directly send the response packet 160 to the client 110.

That is, in the present invention, the receiving server 120 is dedicated to receiving data from the client 110, and the transmitting server 130 is dedicated to transmitting data from the client 110. As described above, in the present invention, data reception and data transmission from the client are separately performed by the reception server 120 and the light transmission server 130, thereby preventing bottlenecks at one point in the overall network service. To this end, the receiving server 120 and the sending server 130 are physically separated and connected to the network.

In addition, the reception server 120 generates the first information 150 corresponding to the service request packet 140 received from the client 110 and transmits the first information 150 to the transmission server 130. The first information 150 includes an IP address and port of the client 110, a port number of the transmission / reception network 100, and the like. As such, the reception server 120 processes the information of the client 110 included in the service request packet 140 received from the client 110 to generate the first information 150, thereby transmitting the server 130. To send. Accordingly, the transmission server 130 may receive the first information 150 and transmit a response packet 160 corresponding to the service required by the client 110 to the client 110. For example, when a service request for downloading data from the client 110 to the server side occurs, the receiving server 120 having the representative IP of the transmission / reception network 100 receives the service request packet 140 from the client 110. After receiving and processing the service request packet 140 to generate first information, the first information 150 is transmitted to the transmission server 130.

The transmitting server 130 receives the first information 150 from the receiving server 120, generates a first packet 160, and transmits the first information 150 to the client 110 in response thereto. The first packet 160 is data for enabling a service required by the client 110. Accordingly, the client 110 may receive the first packet 160 from the transmitting server 130 and receive the required service. For example, when a service request for downloading data from the client 110 to the server side occurs, the transmitting server 130 receives the first information 150 from the receiving server 120, and the transmitting server 130 receives the above information. Based on the first information 150, the data 110, that is, the first packet 160 is transmitted to the client 110. In the first packet 160 transmitted from the transmission server 130 to the client 110, a representative IP address of the transmission / reception network 100 is set as a source network address, and an IP address of the client 110 is set as a destination network address. The port number of the transmission / reception network 100 is set as the source port number, the port number of the client 110 is set as the destination port number, and transmitted to the client 110.

At this time, the transmitting server 130 only transmits the first packet 160 corresponding to the service request packet 140 from the client 110 to the client 110, and does not receive any packet from the client 110. . This may be performed by allowing the receiving server 120 and the sending server 130 to externally use the same representative IP address, but to receive packets received from the outside with the representative IP address only to the receiving server 120. That is, since the server using the representative IP address externally is set as the receiving server 120, a packet having the representative IP address as the destination address is received only by the receiving server 120 by the operation of the TCP / IP network.

The client 110 may receive the first service by receiving the first packet 160 from the transmission server 130. In this case, since the client 110 receives the first packet 160 for the service request packet 140 without any recognition of the separation between the receiving server 120 and the transmitting server 130, the client 110 may receive the first packet 160. The program of can be carried out according to the present invention without any modification.

The client 110 receives the first packet 160 from the transmission server 130 and transmits a second packet (not shown) to the transmit / receive network 100 in response thereto. In this case, the second packet (not shown) is transmitted to the transmission / reception network 100 in response to the first packet 160 received by the client 100 from the transmission server 130. The client 110 transmits the second packet (not shown) to the transmission / reception network 100 to inform that the first packet 160 is normally received from the transmission / reception network 100 side.

As such, the receiving server 120 is responsible for receiving the service request packet 140 from the client 110, and the transmitting server 130 is responsible for transmitting the first packet 160 from the client 110. The transmission server 130 may process the service request of the client 110 more quickly by responding to the service request by using the information of the request packet 140 received from the client 110.

2A is a diagram illustrating a service request packet according to the present invention, FIG. 2B is a diagram illustrating first information according to the present invention, and FIG. 2C is a diagram illustrating a first packet according to the present invention.

The service request packet 140 has information about a client 140 requesting a service and information about a receiving server 120 to receive the service request packet 140. As shown in FIG. Network address 201, destination network address 202, source port number 203, destination network address 204, and data fields 205.

The service request packet 140 has an IP address of the client 110 as a source network address 201 and an IP address of a server to receive the service request packet 140 as a destination network address 202. Accordingly, the client 110 transmitting the service request packet 140 includes the destination network address 202 in the service request packet 140 and transmits the source network at the server side that receives the service request packet 140. The client 110 that has made a service request may be identified by referring to the address 201. In addition, the service request packet 140 includes the port number of the client 110 as the source port number 203 and the port number of the receiving server 120 as the destination port number 204. Accordingly, the service request packet 140 is received through the port of the receiving server 120 corresponding to the destination port number 204, and the receiving server 120 refers to the source port number 203 and the service request packet. It can be seen from which port of the client 110 140 has been sent. For example, as shown in FIG. 2A, the client 110 sets the source network address 201 to 88.24.5.6, which is the client's own IP address, and the destination network address 202 to respond to the service request. 157.8.7.6, which is the IP address of the receiving server 120, is set, the source port number 203 is set to 777, which is the port of the client 110 to request service, and the destination port number 204 is set to respond to the service request. The service request packet 140 is generated by setting the port 80 of the receiving server 120, and the generated service request packet 140 is transmitted to the receiving server 120.

The receiving server 120 receives the service request packet 140, the IP address of the client 110 included in the service request packet 140, the IP address of the receiving server 120, and the client 110. Information such as a port number and a port number of the reception server 120 is processed to generate the first information 150 and transmitted to the transmission server 130. As shown in FIG. 2B, the first information 150 includes an IP address 211 of the client, a port number 212 of the client, a port number 213 of the receiving server, and a data field 214. . As such, the reception server 120 generates the first information 150 using the information included in the service request packet 140 received from the client 110 and transmits the first information 150 to the transmission server 130. For example, if the client IP address 201 included in the service request packet 140 is 88.24.5.6, the client port number 203 is 777, and the receiving server port number 204 is 80, The receiving server 130 uses the information included in the service request packet 140, and the client IP address 211 is 88.24.5.6, the client port number 212 is 777, and the port 213 of the receiving server. Is 80 and the first information 150 including the data field 214 is generated and transmitted to the transmission server 130.

The transmitting server 130 receives the first information 150 from the receiving server 120 through a network connected with the receiving server 120. The first information 150 received by the sending server 130 from the receiving server 120 includes an IP address 211 of the client, a port number 212 of the client, a port number 213 of the receiving server, and the like. By using this, the first packet 160 to be transmitted to the client 110 may be generated. As shown in FIG. 2C, the first packet 160 consists of a source network address 221, a destination network address 222, a source port number 223, a destination port number 224, and a data field. For example, the transmission server 130 sets 157.8.7.6, which is the IP address of the reception server 120, in the source network address 221 using the first information 150 received from the reception server 120. In the destination network address 222, the first packet 160 is generated by setting 88.24.5.6, which is the IP address of the client 110 to receive the first packet 160. In addition, the source server 130 sets the source port number 223, which is the port of the receiving server 120, and the destination port number 224, the port of the client 110 to receive the first packet 160. 777 is set and the first packet 160 is generated together with the data field 225. The transmitting server 130 transmits the first packet 160 to the client 110 that has made a service request, and the client 110 corresponds to the first packet 160, that is, the service request from the transmitting server 130. Since the data is received, the client 110 may receive the service requested.

Thereafter, when the reception of the first packet 160 is normally completed, the client 110 transmits and receives a second packet (not shown) indicating that the first packet 160 has been received from the transmission server 130. Receiving server 120 that transmits to the (100) side, the representative IP of the transmission and reception network 100 receives the second packet (not shown), the service required by the client 110 has been provided normally can confirm.

In the request for the service between the transmission and reception network 110 and the client 110 configured as described above and the provision of the corresponding service, the network connection state between the client 110 and the reception server 120 and the transmission server 130. The configuration and method for checking the network connection state between the client and the client 110 and controlling the transmission / reception rate of the packet according to the network connection state are as follows.

3 is a diagram illustrating a configuration of a distributed flow control system according to another example of the present invention.

As shown in FIG. 3, when the client 310 transmits a service request packet (not shown) to the receiving server 320, the receiving server 320 receives the service request packet (not shown) and the service. First information (not shown) is generated using the request packet (not shown) and transmitted to the transmission server 330.

The transmitting server 330 receives the first information (not shown) from the receiving server 320 and generates a first packet 335 as a response to the receiving server 320 and transmits the first packet 335 to the client 310. The first packet 335 is data corresponding to a service required by the client 310. In the first packet 335 transmitted from the transmission server 320 to the client 310, a representative IP address of the transmission / reception network 300 is set as a source network address, and an IP address of the client 310 is set as a destination network address. The port number of the transmission / reception network 300 is set as the source port number, and the port number of the client 310 is set as the destination port number and transmitted to the client 310. In this case, the transmission server 330 transmits the first packet 335 including a sequence number 340. The sequence number 340 is information on the order of packets transmitted from the transmission server 330 to the client 310.

Since the transmitting server 330 includes the sequence number 340 in the first packet 335 transmitted to the client 310, the transmitting server 330 receives the first packet 335 from the client 310. By using the sequence number 340, the order in which the first packet 335 is transmitted from the transmission server 330 may be known. In addition, the sequence number 340 is used to reconstruct the data corresponding to the service request of the client 310 by arranging the first packets 335 in the order in the client 310. In particular, the sequence number 340 is also used to determine whether the first packet 335 transmitted from the transmission server 330 has been received to the client 310 without packet loss. For example, when the network connection state between the client 310 and the transmission server 330 is not good or a network delay occurs, a service packet for the service request transmitted from the transmission server 330, that is, the first packet is transmitted. There is a case where one packet 335 is not received to the client 310 side. In this case, the sequence number 340 may be checked to determine whether the first packet 335 is received from the transmission server 330.

The client 310 may receive a desired service by receiving the first packet 335 for the service request from the transmitting server 330, and receives the sequence number 340 together with the first packet 335. In addition, the client 310 transmits a second packet 350 informing the reception and reception of the first packet 335 to the transmit / receive network 300, wherein the second packet 350 is transmitted by the client 310 to the transmission server 330. The first packet 335 is received from the packet 335 to indicate that the service required by the client 110 has been normally provided.

The client 310 transmits the second packet 350, which is a response packet informing the reception server 330 of the reception of the first packet 335, to an on-demand method and a periodic method. There are two methods. The on-demand method transmits a response (that is, the second packet 350) every time a service packet (ie, the first packet 335) arrives or every time several service packets 335 arrives. In the periodic method, even if the service packet 335 does not arrive, the response packet 350 is transmitted, and after a predetermined time, the response packet 350 for the current situation is transmitted. On-demand method has an advantage that the response speed is fast because the response is made each time the service packet 335 arrives. However, in the on-demand method, since there is no response when the service packet 335 does not arrive, the reception server 320 cannot know the latest reception status of the client 310, and the response packet 350 is lost. No action can be taken. Since the periodic method transmits the reception status of the client 310 every predetermined time independently of the arrival of the service packet 335, even if the response speed is slow, the recent reception status of the client 310 without the transmission of the service packet 335 Even if the response packet 350 is lost, since the response packet 350 is transmitted again after a while, the reception status of the client 310 can be known even if the service server 335 is not present at the receiving server 320. . According to an embodiment of the present invention, the client 310 uses a mixture of the on-demand method and the periodical method. According to this embodiment, the response speed is also increased, and the reception server 320 can know the reception status of the client 310 even without the service packet 335.

The client 310 generates sequence maps 360 and 370 using the sequence number 340 included in the first packet 335 received from the transmission server 330. The sequence maps 360 and 370 refer to a map formed by sequentially arranging at least one or more of the sequence numbers 340. The client 310 generates the sequence maps 360 and 370 by binding the sequence number 340 by a predetermined number, and receives the generated sequence maps 360 and 370 together with the first packet 350. Send to 320.

The reception server 320 having the representative IP of the transmission / reception network 300 receives the second packet 350 including the sequence maps 360 and 370 from the client 310. Among the sequence maps 360 and 370, a first sequence map 360 is generated using the sequence number 340 received by the client 310 from the transmission server 330, so that the reception server 320 generates the first sequence map 360. 1 is used to determine whether the sequence number 340 included in the sequence map 360 is missing. Particularly, when the sequence number 340 is missing from the first sequence map 360, the reception server 320 determines that a delay occurs in the network between the transmission server 330 and the client 310. , Transmits a flow control packet 325 to the transmission server 330 to regulate transmission of the first packet 340 from the transmission server 330 to the client 310, or to the missing sequence number 340. The packet 325 for retransmitting the corresponding first packet 340 may be transmitted to the transmission server 330.

An example in which the client 310 generates the sequence maps 360 and 370 is as follows. The client 310 receives the first packet 335 including the sequence number 340 from the transmitting server 330, and the first packet including the sequence numbers 340 corresponding to the sequence maps 360 and 370. 335) whether or not it is received. That is, when the client 310 generates a sequence map 350 composed of five sequence numbers 360 such as 1, 2, 3, 4, 5, etc., the client 310 generates the sequence map 350 of the sequence map 350. If the sequence number 360 of 3 and 5 is missed by checking the missing sequence number 360 in step 340, the client 310 checks that the 3 and 5 sequence numbers 360 are missing (1, 2, X). , 4, X generates a sequence map 360 and transmits it to the receiving server 320. The receiving server 320 checks the sequence map 360 received from the client 350 and if it is determined that the 3 and 5 sequence numbers 360 are missing, the receiving server 320 checks the network between the transmitting server 330 and the client 310. Determining that a delay has occurred, a flow control packet 325 is transmitted to the transmission server 330 to regulate transmission of the first packet 335 from the transmission server 330 to the client 310. In addition, the reception server 320 transmits a packet 325 to the transmission server 330 to retransmit the first packet 335 corresponding to the missing sequence numbers 360 and 3. The transmitting server 330 receives the flow control packet 325 from the receiving server 320, and adjusts the transmission rate of the first packet 335 to the client 310 according to the flow control packet 325. do. In addition, the transmission server 330 retransmits the first packet 335 corresponding to 3 and 5, which are the missing sequence numbers 360. Then, the client 310 receives the first packet 340 whose transmission speed is adjusted according to the network between the transmission server 330 and the client 310, and is the missing sequence number 360. The first packet 335 corresponding to 3 and 5 is received again.

Among the sequence maps 360 and 370, the first sequence map 360 is a sequence number 340 for the nk th first packet 335 received from the transmitting server 330, as shown in FIG. 4A. ) Up to the sequence number 340 for the n th first packet 335, and the second sequence map 370 may be configured from the transmission server 330 as shown in FIG. And a sequence number 340 for the received nxk th first packet 335 to a sequence number 340 for the nx th first packet 335. In this case, n, x, and k is an integer of 1 or more. The change of the network delay between the client 310 and the transmission server 330 may be determined by comparing the first sequence map 360 and the second sequence map 370 configured as described above. For example, when n is set to 100, x is 1, and k is 9, the first sequence map 360 is configured as shown in FIG. 370 is configured as disclosed in FIG. 5B. At this time, the first sequence map 360 of the second packet 350 received in time by the receiving server 320 is composed of sequence numbers 340 such as 91, 92, 93, 99, 100, etc. The second sequence map 370 includes sequence numbers 340 such as 90, 91, 92, ... 98, 99, and the like. In addition, the first sequence map 360 of the second packet 350 received later in time by the receiving server 320 includes a sequence number 340 of 92, 93, 94, ... 100, 101, etc. The second sequence map 370 includes sequence numbers 340 such as 91, 92, 93, 99, 100, and the like.

The receiving server 320 receives and compares the sequence maps 360 and 370. That is, the first sequence map 360 of the second packet 350 received in time first and the second sequence map 370 of the second packet 350 received later in time by the receiving server 320 are compared. do. As a result of comparison, when the sequence numbers 340 of 91, 92, 93, 99, 100, etc. match as shown in FIG. 5 (a), the receiving server 320 receives the client 310 from the receiving server 310. It is determined that the network between the servers 320 operates without delay. At this time, the receiving server 320 checks the sequence number 340 of the first sequence map 360 and the second sequence map 370, as shown in (c) of Figure 5, a specific sequence number If 340 is missing, the receiving server 320 sends the flow control packet 325 to the sending server 330, so that the sending server 330 sends the first packet 335 to the client 310. The transmission rate may be adjusted, or the transmission server 330 may retransmit the first packet 335 corresponding to the missing sequence number 340 to the client 310. In particular, when k is set to 1, a more detailed change of the network state between the transmitting server 330 and the client 310 is possible as described above.

In this embodiment, the sequence map is configured by omitting sequence numbers that the client 310 has not received from the transmitting server 330, but the present invention is not limited thereto. For example, a predetermined value (for example, -1) may be displayed at a corresponding position of the sequence map to indicate that a service packet has not been received.

Meanwhile, although the first sequence map 360 is composed of sequence numbers 340 such as 91, 92, 93, 99, 100, etc., the first sequence map 360 is the next sequence map received. 5, the second sequence map 370 is composed of sequence numbers 340 such as 80, 81, 82, ... 88, 89 and received by the receiving server 320. In this case, the receiving server 320 determines that a delay has occurred in the network between the client 310 and the receiving server 320 and thus has not received the sequence maps 360 and 370 sequentially. That is, part of the sequence map that the receiving server 320 should have received is not received. Accordingly, the flow control packet 325 is transmitted to the transmission server 330 to slowly transmit the service request packet 140 from the client 310 to the reception server 320. The transmitting server 330 receives the flow control packet 325 from the receiving server 320 and transmits the packet accordingly to the client 310 so that the client 310 increases the transmission rate of the service request packet 140. Make it adjustable.

6 is a diagram illustrating a receiving server according to an example of the present invention.

As shown in FIG. 6, the reception server 400 according to the present invention includes a network receiver 410, a controller 420, and a network transmitter 430.

The network receiver 410 in the reception server 400 having the representative IP address of the transmission / reception network 300 receives a packet having the representative IP address of the transmission / reception network 300 as a destination network address. Therefore, all packets transmitted to the transmission and reception network 300 are received by the network receiver 410. Therefore, the network receiver 410 receives the service request packet 140 transmitted from the client 310.

In addition, the source network address of the packet transmitted from the transmission server 330 has a representative IP address of the transmission and reception network 300, that is, the IP address of the reception server 320 as the source network address.

The network receiver 610 receives the first sequence map 360 and the second sequence map 370 included in the second packet 350 from the client 310. The first sequence map 360 and the second sequence map 370 are sequence maps generated by the client 310 using the sequence number 340 received from the transmission server 330.

If a specific sequence number 340 is missing from the first sequence map 360 and the second sequence map 370, the controller 620 delays the network between the transmitting server 330 and the client 310. It is assumed that this occurs. Alternatively, the network delay between the client 310 and the receiving server 330 may be calculated by comparing the first sequence map 360 and the second sequence map 370, or the first sequence map 360 and the first sequence map 360 may be calculated. Stores the last sequence map (not shown) most recently received in time than the second sequence map 370, and stores the first sequence map 360 and / or the second sequence map 370 and the final sequence map (not shown). Network delay between the client 310 and the receiving server 330 may be calculated. According to one embodiment of the invention, the network between the client 310 and the receiving server 330 by comparing the first sequence map 360 or / and the second sequence map 370 and the final sequence map (not shown) In addition to the delay, a network transmission error between the client 310 and the receiving server 330 may be determined. For example, the last sequence map that the receiving server 330 receives and stores most recently is composed of sequence numbers of 91, 92, 93, ... 100, and the newly received sequence map is 92, 93, ... should be composed of 101 sequence numbers. If the received sequence map is actually composed of 93, 94, ... 101, 102 sequence numbers, it will consist of 92, 93, ... 101 sequence numbers. It can be determined that a transmission error has occurred in the sequence map. In this case, one or more sequence maps transmitted from the client 310 to the receiving server 330 may be provided.

The network transmitter 630 may recognize the flow control packet 325 for adjusting the transmission of the first packet 340 from the transmission server 330 to the client 310 according to the determination of the network delay of the controller 620. The flow control packet 325 for transmitting to the transmission server 330 or retransmitting the first packet 340 corresponding to the missing specific sequence number 340 may be transmitted to the transmission server 330.

For example, the first sequence map 360 of the second packet 350 received in time by the network receiver 610 may be a sequence number 340 such as 91, 92, 93, 99, 100, or the like. The second sequence map 370 includes sequence numbers 340 such as 90, 91, 92, ... 98, 99, and the like. In addition, the first sequence map 360 of the second packet 350 received later in time is composed of sequence numbers 340 such as 92, 93, 94, ... 100, 101, and the like. 370 includes sequence numbers 340 such as 91, 92, 93, ... 99, 100, and so on.

The controller 620 receives and compares the sequence maps 360 and 370 received by the network receiver 610. When n is set to 100, x is set to 1, and k is set to 9, the first sequence map 360 of the second packet 350 received in time is 91, 92, 93, 99, 100, etc. The second sequence map 370 of the second packet 350 received later in time is identical to the sequence number 340 of 91, 92, 93, 99, 100, etc. ), The controller 620 determines that the network between the client 310 and the receiving server 320 operates without delay. At this time, the receiving server 320 checks the sequence number 340 of the first sequence map 360 and the second sequence map 370, respectively, and if a specific sequence number 340 is missing, the network The transmitter 630 transmits the flow control packet 325 so that the transmission server 330 adjusts the transmission speed of the first packet 335 to the client 310, or the transmission server 330 sends the client ( 310 may retransmit the first packet 335 corresponding to the missing sequence number 340. In particular, when x is set to 1, a more detailed change of the network state between the transmitting server 330 and the client 310 is possible as described above.

7 is a flowchart illustrating a distributed flow control system according to an example of the present invention.

Referring to FIG. 7, the distributed flow control method according to the present invention will be described.

The transmission server 330 transmits the sequence number 340 together with the first packet 335 (S710).

The client 310 receives the sequence number 340, generates the first sequence map 360 and the second sequence map 370 using the received sequence number 340, and then sends the received sequence number to the receiving server 320. The generated sequence maps 360 and 370 are transmitted (S720).

The receiving server 320 receives the first sequence map 360 and the second sequence map 370 from the client 310. Compare the first sequence map 360 of the packets received first temporally and the second sequence map 370 of the packets received later temporally, or the sequence numbers missing from the two sequence maps 360, 370 ( Check 340 respectively. The first sequence map 360 and the second sequence map 370 are compared, and the client 310 and the receiving server according to whether the first sequence map 360 and the second sequence map 370 match. It may be determined whether a delay has occurred in the network between 320 and 320. In addition, it is checked whether a specific sequence number 340 is missing in the first sequence map 360 and the second sequence map 370 to determine whether there is a delay in the network between the client 310 and the transmission server 330. You can judge whether or not.

The reception server 320 determines whether there is a delay in the network between the client 310 and the reception server 320 as described above, and a delay occurs in the network between the client 310 and the transmission server 330. The flow control packet 325 is generated and transmitted to the transmission server 330 by using the result of the determination of whether or not it is performed (S740). The flow control packet 325 includes a response packet transmitted from the transmission server 330, that is, an instruction to adjust or retransmit the transmission rate of the first packet 335, or the client 310 may request a service request packet 140. ) Is sent to send slowly.

The transmitting server 330 receives the service request packet 140 from the receiving server 320, and in response to a command included in the service request packet 140, the first server 335 transmitted to the client 310. Slowly or quickly adjust the transmission speed of the) or retransmit (S750). In addition, the transmission server 330 may transmit a command to the client 310 side, the client 310 to slowly transmit the service request packet 140 to the receiving server 320 side.

The client 310 receives the first packet 335 whose transmission speed is adjusted from the transmission server 330 or re-receives the first packet 335 (S760). In addition, a command for adjusting the transmission speed of the service request packet 140 from the transmission server 330 to the reception server 320 is received, and the service request packet 140 is adjusted and transmitted to the reception server 320.

In the above embodiment, the case where one or two sequence maps are transmitted from the client to the receiving server has been described, but may be extended to three or more within the scope of the present invention.

According to the present invention, it is possible to solve the bottleneck and efficient data communication by distributing the load concentrated on the server without using load balancers such as L4 and L7 switches.

In addition, according to the present invention, since the receiving server is dedicated to receiving data from the client and the transmitting server is dedicated to transmitting data to the client, a distributed system is provided in which no bottleneck occurs at a specific point. In particular, the conventional L4 switch or the like, as well as the packet to be transmitted all received through the L4 switch, the present invention, the received packet is received only to the receiving server, the transmitted packet is not sent through the receiving server Since it is transmitted directly by the transmission server, the problem of the bottleneck according to the prior art is solved.

In addition, according to the present invention, since the network between the client and the receiving server and the network between the client and the sending server are separated from each other, the network delay between the client and the sending server is calculated using sequence maps composed of sequence numbers, respectively. It is possible to adjust the transmission and reception speed of the network between the network and the receiving server, and the network between the client and the sending server, and by checking the sequence number missing in the sequence map, it is possible to retransmit and receive the unreceived packets. It is possible to transmit data efficiently and stably depending on the state of.

In addition, the present invention enables network virtualization that recognizes and uses distributed network resources as one network resource.

As described above, the present invention has been described by way of limited embodiments and drawings, but the present invention is not limited to the above-described embodiments, which can be variously modified and modified by those skilled in the art to which the present invention pertains. Modifications are possible. Accordingly, the spirit of the present invention should be understood only by the claims set forth below, and all equivalent or equivalent modifications thereof will belong to the scope of the present invention.

Claims (19)

A transmission server for generating a first packet and a sequence number associated with the first packet and transmitting the same to a client; And A receiving server that receives from the client a second packet comprising a first sequence map generated by the client using the sequence number Distributed flow control system comprising a. The method of claim 1, wherein the first sequence map, Distributed sequence control system characterized in that the generated by binding the sequence number by a predetermined number. The method of claim 1, wherein the receiving server, Determining whether a sequence number included in the first sequence map is missing, and transmitting a flow control packet including a command to control packet transmission from the transmission server to the client to the transmission server. Distributed flow control system. The method of claim 3, wherein the receiving server, A flow control including instructions for storing a most recently received last sequence map and comparing the first sequence map and the last sequence map received from the client to regulate packet transmission from the client to the receiving server Distributed flow control system for transmitting a packet to the transmission server. The method of claim 1, In response to the service request received from the client to the sending server, the sending server generates a reply packet for the service request packet and transmits the reply packet to the client, and the destination network address of the reply packet is the service. The source network address of the request packet, the source port number of the response packet is the destination port number of the service request packet, and the destination port number of the response packet is the source port number of the service request packet Distributed flow control system. A network receiver configured to receive a packet from the client, the packet including a first sequence map generated by the client using a sequence number received from a transmission server; And A controller for determining whether a sequence number included in the first sequence map is missing and calculating a network delay between the transmitting server and the client Receiving server comprising a. The method of claim 6, A network transmitter that transmits a flow control packet including a command to adjust packet transmission from the transmission server to the client according to the calculated network delay between the transmission server and the client. Receiving server further comprises. The method of claim 6, wherein the control unit, And storing the most recently received last sequence map and comparing the first sequence map and the last sequence map received from the client to calculate a network delay between the client and the receiving server. The method of claim 8, A network transmitter that transmits a flow control packet including a command to control packet transmission from the client to the receiver server according to the determination result of the controller Receiving server further comprises. The method of claim 6, wherein the control unit, A receiving server, wherein the receiving server stores the last sequence map received most recently and compares the first sequence map and the final sequence map received from the client to determine a network transmission error between the client and the receiving server. . The method of claim 6, The first sequence map includes a sequence number for the nkth service packet (n and k are integers greater than 1) received from the transmitting server to a sequence number for the nth service packet. . The method of claim 11, wherein the network receiver, Receiving a second sequence map generated by the client including a sequence number for an nxk-th service packet (x is an integer greater than or equal to 1) received from the transmission server to a sequence number for the nx-th service packet. Receiving server. The method of claim 12, wherein the control unit, And comparing the first sequence map and the second sequence map to determine a change in network delay between the client and the sending server. The method of claim 6, A packet having a destination network address as a representative network address assigned to the sending server and the receiving server is received in the network receiving unit, and the source network address of the first packet received by the client from the transmitting server is the representative network address. A receiving server, characterized in that. The method of claim 6, And the network receiving unit receives a service request packet from the client, and the network transmitting unit transmits a source network address, a source port number, and a destination port number of the service request packet to the transmitting server. The method of claim 6, And the sending server is physically separated from the receiving server and connected to a network. Generating a first sequence map using a sequence number for an n-th service packet from a sequence number for an n-k th service packet received from a transmitting server (where n and x are integers of 1 or more); And Transmitting the first sequence map to a receiving server that is physically separate from the transmitting server. Distributed flow control method comprising a. The method of claim 17, A second sequence map is generated using a sequence number for an nxkth service packet (where k is an integer equal to or greater than 1) received from the transmitting server to a sequence number for an nxth service packet. Transmitting to the receiving server with a first sequence map Distributed flow control method further comprising. A computer-readable recording medium having recorded thereon a program for executing the method of claim 17 or 18.

Images