KR101503717B1 - Network Apparatus and Data Transporting Method Used in the Same - Google Patents

Abstract

The present invention relates to an apparatus and method for transmitting and receiving packets, comprising at least one end system for transmitting and receiving packets, a virtual switch connected to the end system, an uplink port being a physical port connected to the virtual switch, A control agent for managing network information including a MAC address of a link port, at least two hosts located in a cluster, a controller for storing and controlling network information of the host, and at least two networks Wherein the virtual switch establishes a multipath by at least two tunnels through the virtual switch and the network switch located in another host, and transmits and receives the packet.
According to another aspect of the present invention, there is provided a method for transmitting and receiving packet data between at least one end system for transmitting and receiving packets and at least two hosts including the virtual switch and the uplink port, And forming a multi-path between at least two tunnels through a network switch to transmit / receive the packet.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a network device and a data transmission /

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a network device capable of transmitting data using a multi-path by a tunnel among hosts located in a cluster, and a data transmission / reception method used therefor.

As the age of large data has come, network devices have required massive data processing, and as a result of introducing a distributed system to handle them, network performance between horizontal layers has become a big issue. In particular, systems that handle large amounts of data, such as NoSQL, systems that use distributed file systems, or systems that use cloud platforms such as open stacks or cloud stacks, . Therefore, the above systems have become important factors for determining the performance of the entire system in terms of performance of supporting large capacity networks among internal systems. In OSI L2 layer, there are technologies for transmitting and receiving large amount of data through multipath. Typical examples are TRILL standardized by IETF and FabricPath developed by Cisco.

 However, since the above-described technology is a protocol between network switches, it is necessary to use expensive dedicated equipment in order to implement the protocol. Therefore, the cost of constructing the entire system increases and data can be transmitted / received only through dedicated paths (network switches) . Therefore, there is a problem that bottlenecks still occur between the end system and the network switch because data can not be transmitted / received between the end system of the network (for example, a host or a virtual machine) and the network switch through multiple paths.

The present invention provides a network device capable of increasing network bandwidth by forming multiple paths by tunnels between hosts located in the same cluster and transmitting / receiving data, and a data transmission / reception method used therefor.

The network device of the present invention includes at least one end system for transmitting and receiving packets, a virtual switch connected to the end system, an uplink port being a physical port connected to the virtual switch, And a control agent for managing network information including a MAC address of the uplink port, wherein the control agent includes at least two hosts located in a cluster, a controller for storing and controlling network information of the host, Wherein the virtual switch forms a multipath by at least two tunnels through the virtual switch and the network switch located in another host and transmits and receives the packet. At this time, the end system may be composed of the host itself or may include at least one virtual machine.

In addition, the uplink ports located in the host may be formed in a number corresponding to at least the number of the network switches.

In addition, the virtual switch may be configured to not flood the packet received at the uplink port to another uplink port connected to the virtual switch.

In addition, the virtual switch may be configured to transmit the packet using any one of the uplink ports when transmitting the packet transmitted from any one of the specific end systems to the outside of the cluster.

Wherein the tunnel is formed of a tunnel of an L2 layer and includes a MAC address of an uplink port of a transmission virtual switch connected to a transmission end system for transmitting the packet and an uplink port of a receiving virtual switch connected to a receiving end system for receiving the packet, Lt; / RTI > MAC address of the port.

The virtual switch may be configured to transmit the packet transmitted by the transmission terminal system by wrapping the MAC address of the uplink port of the transmitting virtual switch and the MAC address of the uplink port of the receiving virtual switch.

In addition, in the case of transmitting a unicast packet from the transmission terminal system to the reception terminal system located inside the cluster, the transmission virtual switch connected to the transmission terminal system is connected to the reception virtual switch And transmits and receives the packet received from the transmission terminal system by using an extension header formed by a MAC address of an uplink port of the transmission virtual switch and a reception virtual switch and using the tunnel, And remove the extension header from the packet before sending the packet to the receiving end system.

In addition, when transmitting a unicast packet from the transmission terminal system to the outside of the cluster, the transmission virtual switch uses the transmission destination MAC address of the packet received from the transmission terminal system to transmit the packet to the uplink port And to forward the packet to the receiving end system located outside the cluster through the uplink port. At this time, the transmission virtual switch can transmit the packet without using the tunnel.

In addition, when transmitting a multicast packet or a broadcast packet from the transmitting end system to the receiving end system located inside the cluster, the transmitting virtual switch determines whether the destination MAC address of the packet is a multicast address or broadcast Address, the transmission virtual switch confirms the MAC address of the uplink port of the receiving virtual switch to which the receiving end system located in the cluster is connected, and transmits the packet to the transmitting virtual switch and the receiving virtual switch Wherein the virtual switch is configured to transmit the packet to the reception end system after removing the extension header from the received packet by wrapping the packet in an extension header formed by the MAC address of the uplink port of the virtual switch and using the tunnel, .

In addition, when transmitting a multicast packet or a broadcast packet from the transmitting end system to the outside of the cluster, the transmitting virtual switch determines whether the destination MAC address of the packet is a multicast address or a broadcast address, Is located outside the cluster, and transmits the packet using the uplink port allocated to the transmission terminal system. At this time, if the receiving virtual switch located inside the cluster determines that the transmitting end system is located within the cluster from the destination MAC address of the packet to be transmitted, discards the packet, May be formed so as not to be redundantly received by the receiving end system located in the receiving end system.

The receiving virtual switch located in the cluster is further configured to identify the tunnel connected to the uplink port allocated to the transmission terminal system, receive the packet received on the uplink port constituting the tunnel, And discard the packet received via the other uplink port.

In addition, when receiving a multicast packet or a broadcast packet from outside the cluster to a receiving end system located in the cluster, the receiving virtual switch determines whether the destination MAC address of the packet is a multicast or broadcast address, Determining whether the transmission terminal system is located outside the cluster, transmitting the packet received at the uplink port allocated to each reception terminal system to each of the reception terminal systems, May be configured to discard packets received on the uplink port.

The termination system may be connected to the virtual switch by a virtual interface implemented by software.

Also, a data transmission / reception method of the present invention is a method for transmitting / receiving packet type data between at least one end system for transmitting / receiving a packet and at least two hosts including the virtual switch and an uplink port, And forming a multipath by at least two tunnels between the virtual switches to transmit and receive the packets.

In addition, the virtual switch may be configured to not flood the packet received at the uplink port to another uplink port connected to the virtual switch.

In addition, the tunnel is formed of a tunnel of L2 layer, and the MAC address of the uplink port of the transmission virtual switch to which the transmission terminal system for transmitting the packet is connected and the receiving virtual switch And a MAC address of the uplink port.

The virtual switch may be configured to transmit the packet transmitted by the transmission terminal system by wrapping the MAC address of the uplink port of the transmitting virtual switch and the MAC address of the uplink port of the receiving virtual switch have.

The network device and the data transmission / reception method used therefor can form a multipath by a tunnel between hosts located in the same cluster using a virtual switch without using expensive dedicated equipment, Can be increased.

In addition, the network device and the data transmission / reception method used in the present invention increase the network bandwidth between the horizontal layers by multipaths between the layers of the level system, and further increase the network bandwidth between the vertical layers by appropriately distributing the uplink ports of the host There is an effect that can be.

In addition, the network apparatus and the data transmission / reception method used therefor according to the present invention do not generate a network loop or a broadcast storm without using a protocol such as STP, There is an effect that data can be normally transmitted and received through a network in the form of a network.

In the network device and the data transmission / reception method used therefor, when transmitting a multicast packet or a broadcast packet, the MAC address of the destination included in the packet transmitted from the transmission terminal system is checked, When the receiving end system exists in the cluster, the packet is not delivered outside the cluster. The transmitting virtual switch transmits the packet through the multi-path by the tunnel, thereby flooding the packet to the external network switch There is no need to generate a broadcast storm, and it is possible to prevent the learning of the MAC address of the network switch from becoming unstable.

1 is a block diagram of an entire network system to which a network device according to an embodiment of the present invention is connected.
2 is a block diagram of the network device shown in FIG.
FIG. 3 is a block diagram showing a link relationship between a network switch and a tunnel formed between virtual switches in the network device shown in FIG. 1; FIG.
4 is a block diagram illustrating MAC addresses assigned to an end system and an uplink port in the network device shown in FIG.
5 is a flowchart illustrating a method for transmitting a unicast packet within a cluster according to an exemplary embodiment of the present invention.
6 is a flowchart illustrating a method for transmitting a unicast packet to a network device outside a cluster according to an exemplary embodiment of the present invention.
FIG. 7 is a flowchart illustrating a method for transmitting a multicast packet or a broadcast packet in a network according to an exemplary embodiment of the present invention. Referring to FIG.
8 is a flowchart illustrating a method for transmitting a multicast packet or a broadcast packet to a network device outside a cluster according to an exemplary embodiment of the present invention.
9 is a flowchart illustrating a method for transmitting a multicast packet or a broadcast packet to a network device in a cluster according to an exemplary embodiment of the present invention.
10 is a flowchart illustrating a method of receiving a multicast packet or a broadcast packet in which a network device according to an exemplary embodiment of the present invention is transmitted from outside the cluster to the inside.

Hereinafter, a network device according to an embodiment of the present invention and a data transmission / reception method used therein will be described with reference to the accompanying drawings.

First, a configuration of a network device according to an embodiment of the present invention will be described.

1 is a block diagram of an entire network system to which a network device according to an embodiment of the present invention is connected. 2 is a block diagram of the network device shown in FIG. FIG. 3 is a block diagram showing a link relationship between a network switch and a tunnel formed between virtual switches in the network device shown in FIG. 1; FIG. 4 is a block diagram illustrating MAC addresses assigned to an end system and an uplink port in the network device shown in FIG.

1 to 4, a network apparatus according to an embodiment of the present invention includes at least two hosts 100, a controller 200, and at least two network switches 300. The reference numeral 400 denotes a general network to which a network switch is connected.

The host 100 includes at least one end system (ES) 110, a virtual switch 130, at least two uplink ports 140, and a control agent 150.

The host 100 is connected to the network 400 through the network switch 300 and is formed as a general system for transmitting and receiving packets. For example, the host 100 includes an apparatus for connecting to a network and transmitting / receiving data through wired communication, such as a personal computer, a network computer, a mobile terminal, or a multiprocess system. The host 100 may operate as a web server, a file server, a file storage system, or the like.

At least two of the hosts 100 are located in the same cluster. Also, the host 100 includes a virtual switch 130 therein. Here, the cluster means a set of hosts 100 in which the virtual switch 130 is located. The network device forms multipaths by at least two tunnels in the OSI L2 layer between the virtual switches 130 located in each host 100, Data is transmitted and received. In addition, the network device forms a full mesh network between the hosts 100 to transmit and receive data.

Therefore, the network device forms a multipath between the hosts 100 of the same layer by a software method without using expensive dedicated equipment, and transmits / receives the packet.

In addition, the network device transmits and receives data to and from the hosts located outside the same cluster through the network 400.

The termination system 110 refers to a system located at an end in a network, and refers to a system for transmitting and receiving packet-type data through a network. The termination system 110 may be the host 100 itself or at least one virtual machine 120 mounted on the host 100. Thus, the end system 110 may be comprised of only the host 100 or only the virtual machine 120, or may be comprised of the host 100 and the at least one virtual machine 120. Hereinafter, when the host 100 acts as the end system 110, the host 100 is indicated as a separate block and the reference numeral 100a is assigned to the block. On the other hand, when the terminal system 110 transmits data, it refers to the transmission terminal system regardless of whether it is the host 100a or the virtual machine 120, and when receiving the packet, it refers to the reception terminal system.

The termination system 110 includes a virtual switch 130 and at least one termination system internal port 112, 122 for data transmission and reception. The end system internal ports 112 and 122 may be divided into an internal port 112 formed in the host 100a and an internal port 122 formed in the virtual machine 120. [ The end system internal port 112 is a logical internal port and is a virtual interface implemented in software.

The end system 110 is assigned a unique MAC address. For example, if the end system 110 includes one host 100a and N-1 virtual machines 120, 'MAC 1', 'MAC 2', ... , MAC N '. When transmitting the packet, the end system 110 adds its MAC address as a source address and adds the address of the end system 110 to which the packet is to be transmitted as a destination address .

The virtual machine 120 is an application that performs an independent function in the host 100, and can be created as a typical virtual machine. The virtual machine 120 is not created in the host 100 or is generated in at least one host, and the MAC address is assigned to each of the virtual machines 120, like the host 100a. The virtual machine 120 includes at least one virtual machine internal port 122 for transmitting and receiving packet type data to and from the virtual switch 130. The virtual machine internal port 122 is a logical internal port and is a virtual interface implemented in software.

The virtual switch 130 includes a virtual switch internal port 132 for transmitting and receiving data to and from the terminal system 110. The virtual switch 130 is connected to the end system internal ports 112 and 122 of the end system 110 through the virtual switch internal port 132 to transmit and receive packet data. The virtual switch internal port 132 is a logical internal port and is a virtual interface implemented in software.

The virtual switch 130 is connected to the uplink ports 140 and is connected to the network switch 300 which is a physical switch located outside the host 100 through the uplink port 140. The virtual switch 130 is used as an uplink port 140 to form a tunnel of the OSI L2 layer with other virtual switches 130 in the cluster. The uplink port 140 refers to ports of a network interface card (NIC) which is a physical network device installed in the host 100 as described below.

The virtual switch 130 is formed of a software switch, and is located inside each host 100. Therefore, the number of virtual switches 130 may be the same as that of the host 100. The virtual switch 130 forms at least two tunnels t1, t2 and t3 of the OSI L2 layer between the virtual switches 130 and other virtual switches 130 located inside the cluster, ) As a multipath for transmitting and receiving packet type data. That is, the virtual switch 130 is implemented to operate substantially as an L2 layer switch. In addition, when the network device forms a tunnel connecting the L2 layer by the virtual switch 130 in the L3 layer or the L4 layer in addition to the L2 layer, the network device can transmit and receive a packet through a multi-path with a remote network. For example, the network device can configure a virtual network between a data center and a data center to transmit and receive packets in multiple paths.

The virtual switch 130 is connected to the virtual switch 130 of the other hosts 100 located inside the cluster through a network switch 300 by a multipath by a tunnel, Thereby forming a network. For example, referring to FIG. 3, the virtual switch 1 forms three tunnels t1 with the virtual switch 2 through the network switches 1 to 3, and can form three tunnels t2 with the virtual switch 3 . Further, the virtual switch 2 can form the virtual switch 3 and the three tunnels t3 through the network switches 1 to 3. The tunnel is formed of a pair of MAC addresses allocated to each uplink port 140 of the virtual switch 130 located on both sides. Further, the tunnel is formed so as to pass the shortest path between the virtual switches 130. The virtual switch 130 encapsulates the packet in the header of the L2 layer formed by the MAC address of the uplink port 140 and transmits and receives the packet using the multipath. More specifically, the virtual switch 130 is an L2 layer header composed of the MAC address of its own uplink port 140 and the MAC address of the uplink port 140 of the virtual switch 130 receiving the packet Packets are wrapped. Here, the virtual switch 130 receiving the packet means the virtual switch 130 connected to the end system 110 receiving the packet. Meanwhile, the tunnel may be formed by a tunneling protocol (for example, L2TP) used to create a tunnel including an existing protocol.

The virtual switch 130 floods the packet received by the uplink port 140 to another uplink port 140 connected to the virtual switch 130 when receiving the packet using the multi path by the tunnel I never do that. Accordingly, the virtual switch 130 prevents a network loop or a broadcast storm from occurring between the virtual switch 130 and the physical network switch 300. Meanwhile, uplink ports 140 at both ends forming the tunnel are connected through the same network switch 300. In addition, the plurality of tunnels connecting the virtual switch 130 are connected to each other through another network switch 300, thereby increasing the efficiency of transmitting and receiving a packet.

In addition, the virtual switch 130 forms a physical network link with the network switch 300 through the uplink port 140. Accordingly, the virtual switch 130 is located outside the cluster through the network switch 300, and can exchange packet-type data with hosts that do not have the virtual switch 130. In addition, the virtual switch 130 enables data transmission / reception between the hosts 100 located in the cluster and data transmission / reception with the host 100 located outside the cluster. The virtual switch 130 uses any one of the assigned uplink ports 140 when transmitting a packet to the outside of the cluster through the network switch 300 in any one of the specific end systems 110, Port 140 or a plurality of uplink ports 140 are not used. Therefore, the virtual switch 130 prevents the failure of the MAC address learning of the network switch 300.

The uplink port 140 may be a physical input / output port located inside the host 100 and may include input / output ports of a network interface card (NIC). The uplink ports 140 are formed in a number corresponding to at least the number of the network switches 300 and are preferably formed in a number corresponding to the number of the network switches 300. Meanwhile, when the transmission performance of the network switch 300 is higher than that of the uplink port 140, a plurality of uplink ports 140 may be connected to one network switch 300. In this case, the uplink port 140 may be formed in a larger number than the network switch 140. The uplink port 140 is allocated specifically to one or a plurality of the end systems 110 located in the host 100. When the packets are transmitted and received to hosts located outside the cluster, Only the packet transmitted from the end system 110 is received and only the packet received by the assigned end system 110 is received.

Referring to FIG. 3, the uplink port 140 is assigned a unique L2 layer address such as a MAC address. The MAC address of the uplink port 140 is a unique value given by the network interface card manufacturer and is used by the network switch 300 as a physical switch to which the uplink port 140 is connected to improve the address search performance. The sequence number of the corresponding uplink port 140, and other routing information, and is modified or deleted by the control unit 200 and stored.

Also, the uplink port 140 is connected to the network switch 300 located outside the cluster. The uplink ports 140 are preferably connected in a one-to-one manner with different network switches 300, respectively.

The control agent 150 is located inside the virtual switch 130. The control agent 150 is implemented by software and manages network information necessary for data transmission and reception, including the MAC addresses of the end system 110 and the uplink port 140. The network information includes all information necessary to transmit and receive a packet between the host 100 and the network switch 300. For example, the network information may include information on the connection relationship between the end system 110 and the uplink port 140, the MAC address of the end system 110, and the virtual switch 130 to which the end system 110 is connected Information about a path through which packets are transferred between the virtual switches 130, and information about paths through which the virtual switches 130 and packets of the other host 100 are transmitted.

4, information about the end system 110 connected to the virtual switch 130 (table 1), information about the status of the virtual switch 130 connected to another host Information about the multipath formed by the virtual switch 130 included in the uplink port 140 and the uplink port 140 of the virtual switch 130 located at both ends of each multipath, (Table 3), information about the uplink port 140 allocated for communication with the outside of the cluster (Table 4) in each of the end systems 110, and the like. More specifically, with reference to the host 1, Table 1 shows that the virtual switch 1 (Virtual Switch 1) included in the host 1 is connected to the host 100a (MAC 1) and the virtual machines 120 (MAC 3 and MAC 4) And information that the virtual switch 2 included in the host 2 is connected to the host 100a (MAC 2) and the virtual machine 120 (MAC 5, MAC 6), which are the end systems. Table 2 also includes information that the virtual switch 1 is connected to the virtual switch 2 (Virtual Switch 2), which is another virtual switch in the cluster, via paths such as PATH 1 and PATH 2. In Table 3, PATH 1 is formed between the uplink port 1 (MAC 11) of the virtual switch 1 and the uplink port 1 (MAC 21) of the virtual switch 2, and PATH 2 Is formed between the uplink port 2 (MAC 12) of the virtual switch 1 and the uplink port 2 (MAC 22) of the virtual switch 2 (Virtual Switch 2). Table 4 also shows that the host system (MAC 1) and one virtual machine (MAC 3) are connected to the uplink port 1 and that the other virtual machine (MAC 4) is connected to the uplink port 2 .

In addition, the control agent 150 interlocks with the control unit 200, and transmits and receives information to be changed in the respective tables from the control unit 200 to manage the changed information. In addition, the control agent 150 transmits and receives necessary information to and from the control agent 150 located in another host 100.

The control unit 200 includes a database 210 and controls information of a network device including a MAC address of an end system 110 and an uplink port 140 of a host 100 located in a cluster . The control unit 200 may control the network environment such as a MAC address, an IP, a DNS and a routing table necessary for the interface of the end system 110 and the virtual switch 130 of the host 100 located in the cluster, As shown in FIG. That is, the controller 200 stores, changes, or deletes the MAC addresses of the terminal system 110 and the uplink port 140, and manages the MAC addresses. In addition, the controller 200 stores or modifies the information of the internal ports of the end system 110 and the virtual switch 130 to manage the internal ports. The controller 200 transmits and receives information of the network devices including the MAC addresses of the end system 110 and the uplink port 140 to the control agent 150 located in each virtual switch 130.

The controller 200 may be located in any one host 100 located in the cluster and may be located in a plurality of hosts 100 located in the cluster. Also, the controller 200 may be located in a separate server.

The database 210 stores information including the setting information and status of the host 100 located in the cluster. In addition, the database 210 stores information of a network device including MAC addresses of the end system 110 and the uplink port 140 located in the cluster. Here, the end system 110 and the virtual switch 130 include a transmitting side for transmitting a packet and a receiving side for receiving a packet.

The controller 200 is connected to each network switch 300 through a separate control port 220 and transmits and receives information to and from each host 100 connected to the network switch 300.

The network switch 300 is formed of an L2 switch located at an L2 layer of the OSI layer. The L3 switch of the L3 layer or the L4 switch of the L4 layer used in the network device may be used as the network switch 300. [

The network switch 300 determines the connection relationship between the network port 310 of the network switch 300 and the uplink port 140 of the destination of the packet received through the corresponding network port 310, The learned address table is created and managed. For example, referring to FIG. 4, the network switch 1 is configured such that the network port 1 (P 1) is connected to the uplink port 1 (MAC 11) of the virtual switch 1, and the network port 2 A learned address table (table 5) for information connected to the uplink port 1 (MAC 21) of the switch 2 is generated and managed. The network switch 2 and the network switch 3 also generate and manage the learned address table (Table 6, Table 7) for the virtual switch 1 and the uplink port of the virtual switch 2 connected to the respective ports. 4, the information that is deleted from the learned address table managed by the network switch 300 is information learned when the network switch 300 receives the first packet, is not used after the learning, And indicates deleted information after the deletion.

A method for transmitting and receiving packet-type data in a network device according to an embodiment of the present invention will be described in detail.

First, a method in which the network device transmits a unicast packet within a cluster will be described.

5 is a flowchart illustrating a method for transmitting a unicast packet within a cluster according to an exemplary embodiment of the present invention.

An end system ES (hereinafter, referred to as a 'transmission terminal system') 110s for transmitting a packet includes a user data to be transmitted, a MAC address (MAC 1) of the user and an end system ES (Hereinafter, referred to as a "transmission virtual switch") to which the transmission terminal system 110s is connected, including a MAC address (MAC 2) of a reception end system (hereinafter referred to as a "reception end system") 110d, (Virtual Switch 1) 130 (S11). The transmission virtual switch 130s checks whether the receiving end system 110d is an end system located in the same cluster from the MAC address of the receiving end system 110d and checks whether the receiving end system 110d is connected to the virtual switch (Virtual Switch 2) 130d (hereafter referred to as a "received virtual switch"). At this time, the transmission virtual switch 130s obtains information on the MAC address of the receiving virtual switch 130d matching the destination MAC address from which the packet is received from the network information. Also, the transmission virtual switch 130s receives the MAC address (MAC 11, MAC 12, MAC 13) of the uplink port 140s of the transmission virtual switch 130s from the information of the route table managed by the control agent 150 (MAC 21, MAC 22, MAC 23) of the uplink port 140d of the receiving virtual switch 130d.

As described above, the tunnel is formed by a plurality of the MAC addresses of the uplink ports of the transmission virtual switch 130s and the reception virtual switch 130d. In addition, the tunnel is formed using a plurality of network switches 300. For example, the tunnel may include uplink ports 140s of NIC 1 (MAC 11), NIC 2 (MAC 12), and NIC 3 (MAC 13) of the transmission virtual switch 130s and the uplink NIC 1 (MAC 21), NIC 2 (MAC 22), and NIC 3 (MAC 23) which are link ports 140d. The tunnels are also formed through network switches 1, 2 and 3 (300), respectively. The transmission virtual switch 130s transmits an original packet transmitted from the transmission terminal system 110s to a receiving virtual switch 130d receiving the MAC address of the MAC address of its uplink port 140s and a packet, The packet is wrapped with an extension header including the MAC address of the uplink port 140d, which is the MAC address of the uplink port 140d, to form an encapsulated packet (L2 tunnel packet). Meanwhile, since the packet size increases by the size of the extension header, the packet needs to adjust the maximum transmission unit (MTU) value transmitted and received in the transmission terminal system 110s and the reception terminal system 110d do. However, when the network switch 300 supports jumbo frames, the increase of the packet size is not a problem.

The transmission virtual switch 130s transmits a packet (L2 tunnel packet) including an extension header to the network switch 300 (S12). Also, the network switch 300 transmits the received packet to the receiving virtual switch 130d (S13). At this time, the transmission virtual switch 130d applies load balancing to the multipath formed between itself and the receiving virtual switch 130s, thereby improving the overall transmission performance (speed, bandwidth, etc.).

After receiving the packet (L2 tunnel packet), the receiving virtual switch 130d confirms that the transmission destination address and the destination address included in the extension header are the MAC addresses of the uplink ports 140s and 140d constituting the tunnel , It is confirmed that a packet (L2 tunnel packet) is encapsulated in an extension header of the L2 layer and is transmitted in a multi-path. The receiving virtual switch 130d removes the extension header of the L2 tunnel packet and confirms the MAC address MAC2 of the receiving end system 110d and transmits the original packet to the receiving end system 110d (S14).

Next, a method of transmitting the unicast packet from the inside of the cluster to the outside of the cluster by the network device will be described.

6 is a flowchart illustrating a method for transmitting a unicast packet to a network device outside a cluster according to an exemplary embodiment of the present invention.

When the network device transmits a packet to the receiving end system 110d located outside the cluster, there is no virtual switch for receiving the packet at the receiving end system 110d side. Therefore, It is not used. In addition, the network device transmits the packet without encapsulating it.

The transmission terminal system 110s transmits a packet including the data to be transmitted and a MAC address (MAC 3) of the MAC address (MAC 1 or MAC 2) of the receiving end system 110d and the MAC address (MAC 1 or MAC 2) To the switch 130s (S21). The transmission virtual switch 130s checks the MAC address (MAC 3) of the receiving end system 110d to confirm that the receiving end system 110d is not an end system in the cluster. The transmission virtual switch 130s confirms the MAC address of the transmission terminal system 110s and confirms the uplink port (NIC 1 or NIC 2) to transmit the packet to the outside from the uplink port allocation table. At this time, the transmission virtual switch 130s confirms the connection relationship between the transmission terminal system 110s and the uplink ports NIC 1 and NIC 2 by checking from the uplink port allocation table. The transmission virtual switch 130s transmits a packet transmitted from the transmission terminal system 110s to the network switch 300 outside the cluster using only the uplink port (NIC 1 or NIC 2) confirmed from the uplink port assignment table (S22). That is, the transmission virtual switch 130s transmits a packet transmitted from the transmission terminal system 110s of the MAC 1 to the network switch 1 (300) through the uplink port of the NIC 1, And transmits the packet transmitted from the terminal system 110s to the network switch 300 through the uplink port of the NIC2. Accordingly, the network device prevents the address table learned in the network switch 300 from being generated unstably. The transmission virtual switch 130s transmits the packet to the network switch 300 without encapsulating the packet.

The network switch 300 transmits the packet transmitted from the transmission virtual switch 130s to the receiving end system 110d outside the cluster through the network 400 (S23).

Meanwhile, the network switch 300 generates and manages the learned MAC table in the process of transmitting the packet to the receiving end system 110d outside the cluster, and then, from the receiving end system 110d, All of the packets to be transmitted to the transmission terminal system 110s are transmitted to the transmission terminal system 110s through the uplink port of the transmission virtual switch 130s assigned to the transmission terminal system 110s.

Next, a method in which the network device transmits a multicast packet or a broadcast packet in a cluster will be described.

FIG. 7 is a flowchart illustrating a method of transmitting a multicast packet or a broadcast packet within a cluster according to an exemplary embodiment of the present invention.

The transmission terminal system 110s includes a multicast address or a broadcast address OxFFFFFFFFFFFF corresponding to a plurality of reception end systems 110d to which data to be transmitted and its MAC address MAC1 and data are to be transmitted To the transmission virtual switch 1 (130s) (S31). The transmission virtual switch 130s checks whether the destination MAC address of the packet is a multicast address or a broadcast address. Also, the transmission virtual switch 130s confirms all the receiving virtual switches (Virtual Switch 2, ..., Virtual Switch N) 130d to receive a multicast packet or a broadcast packet. The transmission virtual switch 130s repeats the process of transmitting the encapsulated packet (L2 tunnel packet) to each receiving virtual switch 130d in the same manner as the unicast packet transmission process (S32). At this time, the transmission virtual switch 130s transmits the packet through the multipath formed between each virtual switch 130d and each virtual switch 130d.

The receiving virtual switch 130d confirms that the received packet is a packet transmitted through the multipath by confirming that the received packet is encapsulated in the extended header. The reception virtual switch 130d removes the extension header of the packet and confirms the destination MAC address (multicast address or broadcast address) and transmits the packet to the reception end system 110d (S33). The receiving end system 110d receives a packet transmitted from the receiving virtual switch 130d.

Next, a method of transmitting a multicast packet or a broadcast packet from the inside to the outside of the cluster will be described.

8 is a flowchart illustrating a method for transmitting a multicast packet or a broadcast packet to a network device outside a cluster according to an exemplary embodiment of the present invention.

When the network device transmits a multicast packet or a broadcast packet from the inside to the outside of the cluster, the packet is also transmitted to the receiving end system 110di located inside the cluster and the receiving end system 110do located outside the cluster. Therefore, the network device blocks the multicast packet or the broadcast packet from being transmitted to the receiving-end system 110di within the cluster.

The transmission terminal system 110s transmits a multicast address or a broadcast address OxFFFFFFFFFFFF corresponding to a plurality of reception end systems 110di and 110do to transmit data and its own MAC address (MAC 1) To the transmitting virtual switch 130s (S41). The transmission virtual switch 130s checks the destination MAC address of the packet to check whether the destination MAC address is a multicast address or a broadcast address.

Also, the transmission virtual switch 130s checks whether the receiving end system 110di or 110do to transmit a multicast packet or a broadcast packet is located outside the cluster. The transmission virtual switch 130s does not use the multipath when the multicast packet or the broadcast packet is transmitted outside the cluster. After the transmission virtual switch 130s confirms the connection relationship between the transmission terminal system 110s and the uplink port 140s in the uplink port assignment table, the transmission virtual switch 130s uses one uplink port allocated to the transmission terminal system 110s And transmits the packet to the network switch 300 (S42).

Also, since the network switch 300 floods the original packet, the packet is transmitted to other receiving virtual switches (Virtual Switch 2, Virtual Switch 3, ...) 130d located in the cluster (S43) do. In addition, the network switch 130 transmits a packet to the network 400 (S44).

At this time, the receiving virtual switch 130d located inside the cluster receives duplicate packets (S43, S45) from the network switch 300 or the network 400 located outside the cluster (S43, S45) The packet is discarded immediately after the reception of the packet in which the source MAC address is in the cluster in the broadcast packet (S46), so that the packet is not transmitted to the reception end system 110di.

In addition, the receiving end system 110do located outside the cluster receives the packet transmitted through the network 400 (S47).

Next, a description will be given of a method in which the network device transmits a multicast packet or a broadcast packet to and from a cluster.

FIG. 9 is a flowchart illustrating a method for transmitting a multicast packet or a broadcast packet to a network device according to an exemplary embodiment of the present invention. The method according to FIG. 9 is a method for simultaneously performing packet transmission according to FIG. 7 and packet transmission according to FIG.

The transmission terminal system 110s transmits a multicast address or a broadcast address OxFFFFFFFFFFFF corresponding to a plurality of reception termination systems 110di and 110do to transmit data and its own MAC address MAC1 and data, (Virtual switch 1) 130s (S51).

 The transmission virtual switch 130s checks the destination MAC address of the packet to check whether the destination MAC address is a multicast address or a broadcast address. Also, the transmission virtual switch 130s checks whether the receiving end systems 110di and 110do to which packets are to be transmitted are also present outside the cluster. The transmission virtual switch 130s simultaneously transmits the packet to both the inside and outside of the cluster when the receiving end system 110di and 110do to transmit the packet exists both inside and outside the cluster. On the other hand, if the receiving end system 110di is not located outside the cluster, the packet is transmitted according to the method of FIG.

Since the transmission virtual switch 130s does not use a multipath when transmitting a multicast packet or a broadcast packet outside the cluster, the transmission virtual switch 130s transmits the packet according to the method described in FIG. That is, the transmission virtual switch 130s verifies the connection relationship between the transmission terminal system 110s and the uplink port 140s through the uplink port assignment table and transmits the uplink port 140s To the network switch 300 (S52). That is, the transmission virtual switch 130s transmits the multicast packet or the broadcast packet through the uplink port (NIC 1) allocated to the transmission terminal system (MAC 1) without encapsulating it.

In addition, since the network switch 300 floods the original packet, the packet is repeatedly transmitted to other receiving virtual switches (Virtual Switch 2, Virtual Switch 3, ..., ). Also, the network switch 300 transmits a packet through the network 400 (S54).

The receiving virtual switch 130d located inside the cluster duplicates and receives packets from the network switch 300 or the network 400 located outside the cluster at steps S53 and S55. Therefore, the receiving virtual switch 130d confirms the packet in which the source MAC address is located in the cluster in the received multicast packet or the broadcast packet in the route table, and transmits it to the uplink port Only the packet received to the uplink port (NIC 1) 140d of the receiving virtual switch 130d of the tunnel connected to the NIC 1 140s is transmitted to the receiving end system 110di (S56). In addition, the receiving virtual switch 130d discards the packets received by the other uplink ports NIC2 and NIC3 (S57), thereby preventing packets from being transmitted to the receiving end system 110di in the cluster. That is, the Virtual Switch 2 (Virtual Switch 3) 130d confirms the path of the uplink port (NIC 1-MAC 11) 140s allocated to the transmission terminal system 110s, Only the packet transmitted through the uplink port (NIC 1-MAC 21) of the receiving virtual switch 130d is received (S56), and the remaining packets are discarded (S57) do. At this time, the receiving virtual switch 130d checks the uplink port (NIC1-MAC11) of the transmission virtual switch 130s allocated to the transmission terminal system (MAC 1) 110s in the uplink port assignment table. The receiving virtual switch 130d is also connected to the uplink port NIC1-MAC21 of the receiving virtual switch 130d connected to the uplink port (NIC1-MAC11) of the transmitting virtual switch 130s 1-MAC 31), the connection relationship of the tunnel formed between the transmission virtual switch 130s and the reception virtual switch 130d is confirmed.

In addition, the receiving end system 110do located outside the cluster receives the packet transmitted through the network 400 (S58).

Next, a method for receiving a multicast packet or a broadcast packet in which the network device is transmitted from outside the cluster to the inside will be described.

FIG. 10 is a flowchart of an embodiment in which a network device according to an embodiment of the present invention receives a multicast packet or a broadcast packet transmitted into a cluster.

The network device is a multicast packet or a broadcast packet transmitted from the transmission terminal system 110s located outside the cluster to the network 400 in step S61 and received by the network switch 1 300 in step S62, The receiving virtual switch 130d receives a packet to the uplink port NIC 1 140d connected to the network switch 300 in step S63. The receiving virtual switch 130d also receives a packet from another network switch 300 connected to the assigned uplink ports NIC2 and NIC3 (S64).

The receiving virtual switch 130d refers to the uplink port assignment table and transmits a packet received in each uplink port 140d to each receiving end system 130d. That is, the receiving virtual switch 130d transmits the packet received in the uplink port NIC 1 to the receiving-end system 110d having the address of MAC 1 and the receiving-end system 110d of MAC 2 (S65) . In addition, the receiving virtual switch 130d transmits the packet received at the uplink port NIC2 to the receiving end system 110d whose address is MAC 3 (S66). On the other hand, when there is no receiving end system connected to the uplink port (NIC 3) receiving the packet, the receiving virtual switch 130d discards the packet received at the corresponding uplink port (NIC 3) (S67) . The receiving virtual switch 130d checks the connection relationship between the receiving end system 110d and the uplink port 140d through the uplink port assignment table.

100 - Host 110 - Termination system
120 - Virtual machine 130 - Virtual switch
140 - uplink port 150 - control agent
200 - Control unit 300 - Network switch
400 - Network

Claims (20)

At least one end system for transmitting and receiving a packet, a virtual switch connected to the end system, an uplink port being a physical port connected to the virtual switch, and an uplink port connected to the end system and the uplink port And a control agent for managing network information including a MAC address
At least two hosts located inside the cluster
A controller for storing and controlling network information of the host;
At least two network switches connected to the host,
The virtual switch forms a multipath by at least two L2 layer tunnels through the virtual switch and the network switch located in another host and encapsulates the packet in an extension header formed by the MAC address of the uplink port Transmits and receives using the multi-path,
Wherein the end system comprises the host itself or at least one virtual machine, wherein the uplink ports located in the host are formed in a number corresponding to at least the number of the network switches,
Wherein the virtual switch does not flood the packet received at the uplink port to another uplink port connected to the virtual switch. delete delete delete The method according to claim 1,
Wherein the virtual switch is configured to transmit the packet using any one of the specific uplink ports when transmitting the packet transmitted from any one of the specific end systems to the outside of the cluster. The method according to claim 1,
The tunnel is formed of a pair of a MAC address of an uplink port of a transmission virtual switch connected to a transmission terminal system transmitting the packet and a MAC address of an uplink port of a receiving virtual switch connected to a receiving end system receiving the packet The network device comprising: The method according to claim 6,
Wherein the virtual switch is configured to transmit the packet transmitted by the transmission terminal system by wrapping the packet in an extension header formed by the MAC address of the uplink port of the transmission virtual switch and the MAC address of the uplink port of the reception virtual switch, Device. The method according to claim 6,
When transmitting a unicast packet from the transmitting end system to the receiving end system located inside the cluster,
The transmission virtual switch identifies the reception virtual switch connected to the reception terminal system and transmits the packet received from the transmission terminal system to an extension header composed of the MAC address of the transmission virtual switch and the uplink port of the reception virtual switch And transmits and receives using the tunnel,
Wherein the receiving virtual switch removes the extension header from the received packet and then transmits the packet to the receiving end system. The method according to claim 6,
When a unicast packet is transmitted from the transmission terminal system to the outside of the cluster,
Wherein the transmission virtual switch identifies an uplink port to which the packet is to be transmitted using the MAC address of the packet received from the transmission terminal system and transmits the packet through the uplink port to the outside of the cluster To the receiving end system. 10. The method of claim 9,
Wherein the transmission virtual switch transmits the packet without using the tunnel. The method according to claim 6,
When a multicast packet or a broadcast packet is transmitted from the transmission terminal system to the reception terminal system located inside the cluster,
Wherein the transmitting virtual switch connected to the transmitting end system checks whether the destination MAC address of the packet is a multicast address or a broadcast address,
The transmitting virtual switch confirms the MAC address of the uplink port of the receiving virtual switch to which the receiving end system located in the cluster is connected and transmits the packet to the transmitting virtual switch and the receiving virtual switch of the receiving virtual switch MAC address, and transmits the packet using the tunnel,
Wherein the receiving virtual switch removes the extension header from the received packet and then transmits the packet to the receiving end system. The method according to claim 6,
When a multicast packet or a broadcast packet is transmitted from the transmission terminal system to the outside of the cluster,
Wherein the transmitting virtual switch determines whether the destination MAC address of the packet is a multicast address or a broadcast address and whether the receiving end system is also located outside the cluster and transmits the uplink port allocated to the transmitting end system And transmits the packet using the packet. 13. The method of claim 12,
Wherein the receiving virtual switch located in the cluster discards the packet when the transmitting end system is determined to be located in the cluster from the destination MAC address of the packet to be transmitted, So as not to be duplicated in the receiving end system. 13. The method of claim 12,
Wherein the receiving virtual switch located in the cluster identifies the tunnel formed with the uplink port allocated to the transmission terminal system, receives the packet received on the uplink port constituting the tunnel, And discards the packet received via the uplink port. The method according to claim 6,
When receiving a multicast packet or a broadcast packet from an outside of the cluster to a receiving end system located in the cluster,
Wherein the receiving virtual switch determines whether the destination MAC address of the packet is a multicast or broadcast address and whether the transmitting end system is located outside the cluster and transmits the receiving end to the uplink port assigned to the receiving end system To the respective receiving end system, and discards packets received at the uplink port other than the assigned uplink port. delete CLAIMS What is claimed is: 1. A method for transmitting and receiving packets of data between at least one end system for transmitting and receiving packets and at least two hosts including virtual switches and uplink ports,
Forming a multi-path by at least two L2-layered tunnels between the virtual switches, transmitting and receiving the packet by wrapping the packet in an extension header formed by the MAC address of the uplink port,
Wherein the virtual switch does not flood the packet received at the uplink port to another uplink port connected to the virtual switch. delete 18. The method of claim 17,
The tunnel is formed of a pair of a MAC address of an uplink port of a transmission virtual switch to which a transmission terminal system transmitting the packet is connected and a MAC address of an uplink port of a receiving virtual switch to which the receiving end system receiving the packet is connected The data transmission / reception method comprising the steps of: 20. The method of claim 19,
Wherein the virtual switch transfers the packet transmitted by the transmission terminal system with an extension header formed by an MAC address of an uplink port of the transmission virtual switch and an MAC address of an uplink port of the reception virtual switch, Transmitting / receiving method.

Images