WO2013049989A1

WO2013049989A1 - Layer two interconnection between data centers

Info

Publication number: WO2013049989A1
Application number: PCT/CN2012/081575
Authority: WO
Inventors: Wei Li
Original assignee: Hangzhou H3C Technologies Co., Ltd.
Priority date: 2011-10-08
Filing date: 2012-09-19
Publication date: 2013-04-11
Also published as: CN102355417A

Abstract

According to an implementation, in a method, an edge device is enabled with MFF, proxy forwards ARP request and ARP response according to the MFF mechanism, and learns ARP information according to the received ARP response; when receiving a packet, the edge device modifies the source MAC address of the packet to the MAC address of the edge device itself, searches corresponding ARP information according to a destination IP address of the packet, modifies a destination MAC address of the packet according to the searched ARP information, and then forwards the modified packet.

Description

LAYER TWO INTERCONNECTION BETWEEN DATA CENTERS BACKGROUND

[0001 ] The typical internet data center (IDC) provides network connectivity and other services mainly based on the sale of bandwidth and cabinet space and hosting, and provides infrastructure services in the unit of a computer room. With the rapid development of the IT industry, user traffic volume has increased explosively. Operators of I DCs are therefore facing unprecedented difficulties, such as space resource constraints, uneven development of high and low end computer rooms, low utilization of hardware resources, higher construction and operating costs, single business structure, pressure in business competition and technological development, etc.

[0002] One conventional technique to overcoming some of these difficulties is to implement cloud computing services, which has been found to improve equipment utilization, reduce operating costs, and improve profit margins. At present, the principal methods in which cloud computing services are implemented are through "virtualization" and flexible scheduling of resources. Particularly, the flexible scheduling of computing resources may be realized through the virtualization. Within the same data center, the flexible scheduling of computing resources is relatively easily achieved. However, the difficulty lies in the realization of flexible scheduling of resources between a plurality of data centers, which require that a wide ranging layer 2 network be constructed.

[0003] However, the construction of a wide ranging layer 2 network is relatively difficult because layer 2 interconnections between IDC clouds will cause media access control (MAC) addresses of the plurality of IDC clouds to double, which will make switches difficult to load. The construction of a wide ranging layer 2 network will also result in a large range of broadcast domains, e.g., layer 2 interconnections between IDC clouds will result in a large layer 2 broadcast domain formed between a plurality of IDC clouds. In addition, address resolution protocol (ARP) broadcast and unknown unicast will be further spread, which increases the threat of layer 2 network broadcast storms.

[0004] The typical MAC address learning style is a learning style based on source MAC addresses of hardware chips, which may be relatively simply implemented but may not identify and distinguish between messages, and will automatically learn all received messages, thus causing a MAC table to become too large in size.

[0005] At present, for the deficiencies of the MAC address learning of the learning style, which is based on source MAC addresses of hardware chips, the overlay transport virtualization (OTV) technology uses a technique known as "address flooding learning," which is based on the data plane, to control plane protocols: intermediate system to intermediate system (IS-IS) routing protocol to achieve the MAC address learning. In this way, the size of MAC address learning may be controlled, and the problem of oversized MAC addresses may be solved.

BRIEF DESCRIPTION OF THE DRAWINGS

[0006] Features of the present disclosure are illustrated by way of example and not limited in the following figure(s), in which like numerals indicate like elements, in which:

[0007] Fig. 1 is a schematic diagram showing a process of a MAC address learning of an OTV technology, according to an example of the present disclosure.

[0008] Fig. 2 is a schematic diagram of a layer 2 interconnection network between data centers, according to an example of the present disclosure.

[0009] Fig. 3 is a schematic diagram of dividing a data center site shown in Fig. 2 into a plurality of layer 2 broadcast domains, according to an example of the present disclosure.

[0010] Fig. 4 is a flow chart of a method of layer 2 interconnection between data centers, according to an example of the present disclosure.

[0011 ] Fig. 5 is a schematic diagram of an edge device, according to an example of the present disclosure.

DETAILED DESCRIPTION

[0012] For simplicity and illustrative purposes, the present disclosure is described by referring mainly to an example thereof. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present disclosure. It will be readily apparent however, that the present disclosure may be practiced without limitation to these specific details. In other instances, some methods and structures have not been described in detail so as not to unnecessarily obscure the present disclosure.

[0013] Throughout the present disclosure, the terms "a" and "an" are intended to denote at least one of a particular element. As used herein, the term "includes" means includes but not limited to, the term "including" means including but not limited to. The term "based on" means based at least in part on.

[0014] Referring to Fig. 1 , there is shown a schematic diagram of a process of a MAC address learning of an OTV technology, according to an example of the present disclosure. The data center sites A, B, and C access a backbone network through edge devices 1 , 2, 3, respectively. The topologies of each data center site A, B, and C are not drawn. When the edge device of each data center site learns a new MAC address from within the data center site, the edge device will determine the associated VLAN ID and the next hop IP address (an interface IP through which the edge device is connected to the backbone network) through the IS-IS routing protocol. The edge device will also copy and forward the new MAC address to all of the neighbors in the backbone network through multicast trees. For example, as shown in Fig. 1 , the dotted line 11 represents that the edge device 1 copies and forwards a new MAC address learned from within the data center site A to the data center site B through the multicast trees in the backbone network and the dotted line 12 represents that the edge device 1 copies and forwards the new MAC address learned from within the data center site A to the data center site C through the multicast trees in the backbone network.

[0015] However, realization of the OTV technology is too complex because it needs to apply the IS-IS routing protocol in the layer 2 network and make new changes (which relate to modifications of the IETF standards and drastic changes of the software plane and so on) to the IS-IS routing protocol. Further, the complexity of realizing the OTV technology also causes the number of its MAC address learning to be limited. Moreover, the OTV technology is applicable in data center interconnect application of some enterprises, but may not be applied to large-scale data center interconnections, particularly interconnections between clouds of IDC operators.

[001 6] The present disclosure describes an example in which an edge device in a data center site is enabled with MAC forced forwarding (MFF). The edge device proxy forwards an ARP request and an ARP response from within the data center site or from a remote data center site according to the MFF mechanism, and records corresponding ARP information according to the received ARP response. When receiving a packet from within the data center site or from the remote data center site, the edge device modifies and proxy forwards the packet according to the ARP information corresponding to a destination IP address of the packet, thereby realizing the layer 2 interconnection between data centers.

[0017] Here, the MFF achieves ARP proxy responding. Particularly, an Ethernet access node (EAN) intercepts a user equipment's ARP request and takes a MAC address of a designated access router (AR) as a responded MAC address through the ARP proxy responding mechanism, thereby forcing users to send all traffic to the designated AR.

[0018] Referring to Fig. 2, there is shown a schematic diagram of a layer 2 interconnection network between data centers, according to an example of the present disclosure. EAN1 , EAN2, and EAN3 are edge devices of data center sites A, B, and C, respectively. A MAC address of the EAN1 is MAC A, the EAN1 is connected with user equipment within the data center site A through a port P11 and is connected with the backbone network through a port P12. A MAC address of the EAN2 is MAC B, the EAN2 is connected with user equipment within the data center site B through a port P21 , and is connected with the backbone network through a port P22. A MAC address of the EAN3 is MAC_C, the EAN3 is connected with user equipment within the data center site C through a port P31 , and is connected with the backbone network through a port P32. The site A includes user equipment H1 and H2. An IP address and MAC address of the user equipment H1 are IP1 and MAC1 , respectively. An IP address and MAC address of the user equipment H2 are IP2 and MAC2, respectively. The site B includes a user equipment H3. An IP address and MAC address of the user equipment H3 are IP3 and MAC3, respectively. The site C includes a user equipment H4. An IP address and MAC address of the user equipment H4 are IP4 and MAC4, respectively. The EAN1 , EAN2, and EAN3 are enabled with a media access controller (MAC) forced forwarding (MFF). In addition, the EAN1 , EAN2, and EAN3 also carry functions of an AR.

[0019] When the user equipment H1 in the site A wants to communicate with the user equipment H3 in the site B, the user equipment H1 first needs to learn the MAC address of the user equipment H3 and may learn the MAC address through an ARP request.

[0020] The user equipment H1 sends an ARP request of which a destination IP address is IP3. Since the EAN1 is enabled with MFF, after the EAN1 intercepts the ARP request in the port P11 , the EAN1 modifies a source MAC address of the ARP request: MAC1 to the MAC address: MAC A of the EAN1 itself, and sends the modified ARP request to the backbone network through the port P12.

[0021 ] The EAN2 and the EAN3 both may receive the ARP request, but since the user equipment H3 is not in the site C, the EAN3 will not return an ARP response.

[0022] After the EAN2 receives the ARP request in the port P22, since the EAN2 is enabled with MFF, the EAN2 modifies a source MAC address of the ARP request: MAC A to the MAC address: MAC B of the EAN2 itself, and broadcasts the modified ARP request in the site B through the port P21 . After the user equipment H3 receives the ARP request, the user equipment H3 finds that the destination IP address is its own IP address, thus, the user equipment H3 learns the ARP information: IP1 and MAC B carried in the ARP request, and returns an ARP response; a source IP address and source MAC address of the ARP response are IP3 and MAC3, respectively. After the EAN2 receives the ARP response of the user equipment H3 in the port P21 , the EAN2 learns the ARP information: IP3 and MAC3 of the ARP response, and modifies the source MAC address of the ARP response to the MAC address: MAC B of the EAN2 itself, and returns the modified ARP response to the EAN1 through the port P22. After the EAN1 receives the ARP response in the port P12, the EAN1 learns the ARP information: IP3 and MAC B of the ARP response, and modifies the source MAC to the MAC address: MAC A of the EAN1 itself, and returns the modified ARP response to the user equipment H1 through the port P11 . After the user equipment H1 receives the ARP response, the user equipment H1 records the ARP information: IP3 and MAC A of the ARP response. At this point, the user equipment H1 obtains the MAC address of the user equipment H3.

[0023] Here, since the EAN1 is enabled with MFF, the MAC address of the host H3 obtained by the user equipment H1 is not the real MAC address of the user equipment H3, but is the MAC address of the EAN1 .

[0024] After the user equipment H1 obtains the MAC address of the user equipment H3, the user equipment H1 can start to communicate with the user equipment H3, and the specific communication process is as follows: the user equipment H1 send a packet of which a source IP is IP1 , a source MAC address is MAC1 , a destination IP address is IP3, and a destination MAC address is MAC A; after the EAN1 receives the packet in the port P11 , the EAN1 modifies the source MAC address of the packet to MAC_A, searches corresponding ARP information in an ARP cache according to the destination IP address of the packet: IP3, modifies the destination MAC address of the packet to MAC_B according to the searched ARP information: IP3 and MAC B, and then sends the modified packet to the backbone network through the port P12; after the EAN2 receives the packet in the port P22, the EAN2 modifies the source MAC address of the packet to MAC_B, searches corresponding ARP information in the ARP cache according to the destination IP address of the packet: IP3, modifies the destination MAC address of the packet to MAC3 according to the searched ARP information: IP3 and MAC3, and then sends the modified packet to the user equipment H3 through the port P21 ; and after the user equipment H3 receives the packet, the user equipment H3 finds that the destination IP address and the destination MAC address of the packet are its own IP address and MAC address, respectively, thus, the user equipment H3 accepts and normally processes the packet.

[0025] It may be seen, in the above process, that the edge device of each data center site intercepts ARP protocol messages between user equipment according to the MFF mechanism, records corresponding ARP information to modify the source MAC address and the destination MAC address of received packets in the communication process between the user equipment, and proxy forwards modified packets, thereby realizing layer 2 interconnection between data centers.

[0026] In practical applications, in order to reduce unnecessary ARP broadcasts, when an edge device of one data center site receives an ARP request, the edge device may first determine whether ARP information corresponding to the destination IP address of the ARP request already exists. If the ARP information already exists, the MAC address of the edge device itself may be directly returned. For example, in Fig. 2, after the user equipment H1 sends the ARP request of which the destination IP address is IP3 and receives the corresponding ARP response, the EAN1 already records the ARP information: IP3 and MAC B corresponding to the IP3. [0027] At this point, if the user equipment H2 in the site A wants to communicate with the user equipment H3 in the site B, the user equipment H2 may first send a request of which a destination IP address is IP3 to obtain a MAC address of the user equipment H3. After the EAN1 intercepts the ARP request, the EAN1 searches the ARP cache to determine that the ARP information corresponding to the IP3 already exists (e.g. is already stored in the edge device receiving the ARP request). Thus, the MAC address: MAC A of the EAN1 itself is directly returned to the user equipment H2. In this way, an ARP information corresponding to the user equipment H3, which is learned by the user equipment H2 is IP3 and MAC A.

[0028] In this way, the user equipment H2 may send a packet of which the destination IP address is IP3 and the destination MAC address is MAC_A according to the obtained ARP information corresponding to the user equipment H3. After the EAN1 receives the packet, the EAN1 modifies the source MAC of the packet to MAC A, searches corresponding ARP information in the ARP cache according to the IP3, modifies the destination MAC of the packet to MAC_B according to the searched ARP information: IP3 and MAC B, and sends the modified packet to the EAN2. The EAN2 modifies the source MAC address of the packet to MAC B, searches corresponding ARP information: IP3 and MAC3 in the ARP cache according to the IP3, modifies the destination MAC address of the packet to MAC3, and sends the modified packet to the user equipment H3. At this point, the communication between the user equipment H2 and the user equipment H3 is realized.

[0029] Further, for the situation that a corresponding ARP response is not received for a sent ARP request, a punishment mechanism may be established to further reduce unnecessary ARP broadcasts, thereby reducing the number of ARP broadcast messages in the network. For example, as shown in Fig. 3, if the user equipment H1 requests a MAC address of dropped or non-existing user equipment, the EAN1 intercepts the ARP request and then broadcasts it through the backbone network, the EAN2 and the EAN3 receive the ARP request and then broadcast the ARP request within their respective sites. If the user equipment H1 continuously and repeatedly sends the ARP request, it will occupy a large amount of network resources, resulting in unnecessary waste of resources. For this, the EAN1 may establish a punishment record for the ARP request for the user equipment H1 , which sends the ARP request. In addition, if the ARP request or a ARP request sent by the user equipment H1 is received once again within a preset time, the ARP request is directly discarded, thereby reducing unnecessary ARP broadcasts.

[0030] In addition, after the edge devices in the data center sites are enabled with MFF, the edge devices may also support the migration of virtual machines between the data center sites. Before and after the migration of the virtual machines, the IP addresses and MAC addresses of the edge devices remain unchanged. Still taking Fig. 2 as an example, it is assumed that a virtual machine X in the user equipment H1 migrates to the user equipment H3, and an IP address of the virtual machine X is IP1 , and an MAC address is MAC1 . The migration process of the virtual machine X is as follows: after the virtual machine X is restarted on the user equipment H3, the virtual machine X may send a free ARP message of which a source IP address is IP1 and a source MAC address is MAC1 ; the free ARP message is broadcasted in the site B, so that all the switches in the site B perform MAC address learning according to the free ARP message to re-learn the MAC address of the virtual machine X; the EAN2 intercepts the free ARP message, and searches for ARP information corresponding to IP1 in the ARP cache; if the ARP information is found, it shows that the virtual machine X is originally in the site B and does not migrate, thus, it is just needed to update the searched ARP information; if the ARP information is not found, then the EAN2 modifies the source address of the free ARP message to MAC B and send the modified free ARP message to the backbone network; after the EAN1 receives the free ARP message and finds the ARP information corresponding to IP1 in the ARP cache, it shows that the virtual machine X migrates from the site A to the site B, thus, the EAN1 updates the ARP information and modifies the source MAC address of the free ARP message and then broadcasts the modified free ARP message in the site A, so that all the switches in the site A perform MAC address learning according to the received free ARP message to re-learn the MAC address and corresponding output port of the virtual machine; and after the EAN3 receives the free ARP message, the EAN3 does not find the ARP information corresponding to IP1 in the ARP cache, it shows that a migration event in which the virtual machine X migrates from the site C to the site B is not happened, thus, no processing is needed to be performed.

[0031 ] It may be seen from the above migration process that after the migration of the virtual machine, the free ARP message is sent through the user equipment to which the virtual machine migrates. After the edge devices of the data center sites receive the free ARP message, the edge devices of the data center sites identify the ARP message of the virtual machine after the migration through matching ARP information, which may make each user equipment in the data center site in which the virtual machine is located before the migration learn that the virtual machine migration has occurred. Each user equipment in the data center site in which the virtual machine is located before the migration may then update respective corresponding ARP information according to the free ARP message and the migration process of the virtual machine may be completed, thereby realizing rapid migration of the virtual machine. Further, in actual implementation, reverse ARP messages can also be used.

[0032] In the example shown in Fig. 2, the edge device of each data center site is enabled with MFF, proxy responds to the user equipment's ARP request according to the MFF mechanism, learns ARP information based on interactive business in this process, and proxy forwards packets between data center sites according to the learned ARP information, thereby realizing layer 2 interconnection between data centers, and ensuring that MAC addresses within each data center site will not be spread to other data center sites through the backbone network. Through the exchange mechanism within each data center site, the edge device can learn MAC addresses within the data center site in which it is in, and cannot learn MAC addresses of other data center sites, thus the size of the MAC address table utilized by each of the edge devices is not greatly increased. Further, in the above examples, the edge devices learn ARP information just based on interactive business, that is, just when receiving a corresponding ARP response after sending an ARP request, the edge devices learn the ARP information and send the leaned ARP information to hardwires, rather than learn ARP information of all ARP requests. Thus, the ARP table items are also substantially minimized and will not affect the forwarding performance of hardwire connections.

[0033] The layer 2 interconnection between data centers is actually a layer 2 interconnection between broadcast domains constituted by the data center sites. For some massive data center sites, the data center site may also be further divided into a plurality of layer 2 broadcast domains. In this example, each layer 2 broadcast domain is equipped with edge devices that have the same functions as those of the edge devices of the data center sites, thereby achieving multi-level MFF architecture and then reducing the number of MAC addresses in each level of the MFF architecture, further narrowing the broadcasting scope of ARP. The example shown in Fig. 2 actually belongs to multi-level MFF architecture: two-level MFF architecture, when the user equipment between data center sites communicate, an ARP request needs to go through the edge device of the site where it is in proxy and proxy responding for the edge device of the remote site. A packet also needs to go through the edge device of the site where it is in and proxy forwarding for the edge device of the remote site, in order to achieve communication.

[0034] Based on the data center sites shown in Fig. 2 and in combination with Fig. 3, the following will describe the communication situation when the data center sites are divided into a plurality of layer 2 broadcast domains to form a multi-level MFF architecture. [0035] Referring to Fig. 3, there is shown a schematic diagram of dividing the data center site shown in Fig. 2 into a plurality of layer 2 broadcast domains. As shown in Fig. 3, the data center site A is divided into a layer 2 broadcast domain A1 and a layer 2 broadcast domain A2, which are equipped with an edge device EAN11 having a MAC address of MAC_A1 and an edge device EAN12 having a MAC address of MAC_A2, respectively. The layer 2 broadcast domain A1 includes the user equipment H1 , the layer 2 broadcast domain A2 includes the user equipment H2. The EAN11 and the EAN12 have the same functions as those of the EAN1 and the EAN11 . The EAN12 and the EAN1 are connected to the same network through their respective ports.

[0036] When the user equipment H1 wants to communicate with the user equipment H3, the user equipment H1 first obtains a MAC address of the user equipment H3 by sending an ARP request, the process is as follows: the user equipment H1 sends an ARP request of which the destination IP address is IP3; the EAN11 intercepts the ARP request, modifies the source MAC address of the ARP request to MAC A1 according to the MFF mechanism, and sends the modified ARP request; the EAN1 and the EAN12 both may receive the ARP request since the user equipment H3 does not belong to the layer 2 broadcast domain A2, thus, the EAN12 will not return an ARP response; after the EAN1 receives the ARP request, the EAN1 modifies the source MAC of the ARP request to MAC A, and sends the modified ARP request to the backbone network; after the EAN2 receives the ARP request, the EAN2 modifies the source MAC address of the ARP request to MAC_B, and broadcasts the modified ARP request in the site B; after the user equipment H3 receives the ARP request, the user equipment H3 learns the ARP information: IP1 and MAC B, and returns an ARP response of which the source IP address is IP3 and the source MAC address is MAC3; after the EAN2 receives the ARP response, the EAN2 learns the ARP information: IP3 and MAC3, and modifies the source MAC address of the ARP response to MAC B, and returns the modified ARP response to the EAN1 ; after the EAN1 receives the ARP response, the EAN1 learns the ARP information: IP3 and MAC B, and modifies the source MAC of the ARP response to MAC A, and returns the modified ARP response to the EAN11 ; after the EAN11 receives the ARP response, the EAN11 learns the ARP information: IP3 and MAC A, and modifies the source MAC of the ARP response to MAC A1 , and returns the modified ARP response to the user equipment H1 ; after the user equipment H1 receives the ARP response, the user equipment H1 learns the ARP information: IP3 and MAC_A1 .

[0037] Then, the user equipment H1 may start to communicate with the user equipment H3 according to the obtained MAC address of the host H3 (which is actually the MAC address of EAN11 ), and the communication process is as follows: the user equipment H1 sends a packet of which the source IP is IP1 , the source MAC address is MAC1 , the destination IP address is IP3, and the destination MAC address is MAC_A1 ; after the EAN11 receives the packet, the EAN11 modifies the source MAC address of the packet to MAC_A1 , searches corresponding ARP information according to the destination IP address of the packet, modifies the destination MAC address of the packet to MAC_A according to the searched ARP information: IP3 and MAC A, and then sends the modified packet to the EAN1 ; after the EAN1 receives the packet, the EAN1 modifies the source MAC address of the packet to MAC_A, searches corresponding ARP information according to the destination IP address of the packet, modifies the destination MAC address of the packet to MAC_B according to the searched ARP information: IP3 and MAC B, and then sends the modified packet to the EAN2; after the EAN2 receives the packet, the EAN2 modifies the source MAC address of the packet to MAC_B, searches corresponding ARP information according to the destination IP address of the packet, modifies the destination MAC address of the packet to MAC3 according to the searched ARP information: IP3 and MAC3, and then sends the modified packet to the user equipment H3; at this point, the communication between the user equipment H1 and the user equipment H3 is realized. It may be seen that after the data center sites are divided into a plurality of layer 2 broadcast domains to form the multi-level MFF architecture, the normal communication of user equipment in the data center sites may still be ensured, and the broadcasting scope of ARP may be further narrowed.

[0038] Based on the principle of the preceding description, one example of the present disclosure provides a method of layer 2 interconnection between data centers and an edge device.

[0039] Referring to Fig. 4, there is shown a flow chart of a method of layer 2 interconnection between data centers according to an example of the present disclosure. The method includes the following blocks.

[0040] Block 401 : when an edge device receives an ARP request information of a user equipment in a first port, the edge device modifies a source MAC address of the ARP request to an MAC address of the edge device itself, sends the modified ARP request through a second port; after receiving a corresponding ARP response from a second port, the edge device records ARP information, modifies a source MAC address of the ARP response to the MAC address of the edge device itself and sends the modified ARP response through the first port.

[0041 ] Here, the edge device is enabled with MFF in advance, the first port is a port that is connected with user equipment within the site, and the second port is a port that is connected with edge devices of remote sites. [0042] Block 402: when the edge device receives an ARP request of an edge device of a remote site in the second port, the edge device modifies a source MAC address of the ARP request to the MAC address of the edge device itself, sends the modified ARP request through the first port, after receiving a corresponding ARP response from the first port, the edge device records ARP information, modifies a source MAC address of the ARP response to the MAC address of the edge device itself and sends the modified ARP response through the second port.

[0043] Block 403: when the edge device receives a packet of which a destination MAC is the MAC address of the edge device itself from a user equipment in the first port, the edge device modifies a source MAC address of the packet to the MAC address of the edge device itself, modifies the destination MAC address of the packet according to ARP information corresponding to the destination IP of the packet, and sends the modified packet through the second port.

[0044] Here, the edge device discards a packet of which a destination MAC is not the MAC address of the edge device itself.

[0045] Block 404: when the edge device receives a packet of which a destination MAC is the MAC address of the edge device itself from an edge device of a remote site in the second port, the edge device modifies a source MAC address of the packet to the MAC address of the edge device itself, modifies the destination MAC address of the packet according to the ARP information corresponding to the destination IP of the packet, and sends the modified packet through the first port.

[0046] Here, the edge device discards a packet of which a destination MAC is not the MAC address of the edge device itself.

[0047] In the example of the present disclosure shown in Fig. 4, when the current EAN receives an ARP request, if the ARP information corresponding to the destination IP of the ARP request already exists, then it is not needed to proxy forward the ARP request, and a corresponding ARP response may be directly returned, thereby greatly reducing the ARP broadcast in the layer 2 network. [0048] Therefore, after the current EAN receives the user equipment's ARP request in the first port and before the current EAN sends the modified ARP request through the second port after the current EAN has modified the source MAC address of the ARP request to the MAC address of the current edge device itself, the following is further included: the current EAN searches corresponding ARP information in the ARP cache according to the destination IP of the ARP request. If the ARP information is found, then the current EAN returns its own MAC address to the user equipment. If the ARP information is not found, then the current EAN modifies the source MAC address of the ARP request to the MAC address of the current edge device itself and sends the modified ARP request through the second port.

[0049] After the current EAN receives an ARP request from a superior EAN in the second port and before the current EAN sends the modified ARP request through the first port after the current EAN has modified the source MAC address of the ARP request to the MAC address of the current edge device itself, the following is further included: the current EAN searches corresponding ARP information in the ARP cache according to the destination IP of the ARP request. If the ARP information is found, then the current EAN returns its own MAC address to the superior EAN. If the ARP information is not found, then the current EAN modifies the source MAC address of the ARP request to the MAC address of the current EAN and sends the modified ARP request through the first port.

[0050] Further, for the situation that a corresponding ARP response is not received for a sent ARP request, a punishment mechanism may also be established to further reduce ARP broadcasts in the layer 2 network.

[0051 ] A punishment record may be established for the ARP request. In this example, if the ARP request is received once again within a preset time, the ARP request is no longer proxy forwarded. The specific implementation may be as follows:

[0052] After the current EAN modifies the source MAC address of the ARP request to the MAC address of the current EAN and sends the modified ARP request through the second port, the following is further included: if a corresponding ARP response is not received within a preset time, then a punish sign and a punish aging time are set for the ARP request, and the punish sign for the ARP request will be deleted when the aging time is exceeded.

[0053] In this way, after the current EAN receives the user equipment's ARP request in the first port and before the current EAN sends the modified ARP request through the second port after the current EAN has modified the source MAC address of the ARP request to the MAC address of the current EAN, the following may be further included: determining whether there is a punish sign for the ARP request, if yes, the ARP request will not be sent; otherwise, the current EAN modifies the source MAC address of the ARP request to the MAC address of the current EAN and sends the modified ARP request through the second port.

[0054] A punishment record may also be established for a user equipment that sends the ARP request. In this example, if an ARP sent by the user equipment is received once again within a preset time, the ARP request is no longer proxy forwarded. The specific implementation may be as follows:

[0055] After the current EAN modifies the source MAC address of the ARP request to the MAC address of the current EAN and sends the modified ARP request through the second port, the following is further included: if a corresponding ARP response is not received within a preset time, then a punish sign and a punish aging time are set for the user equipment that sent the ARP request, and the punish sign for the user equipment that sent the ARP request will be deleted when the aging time is exceeded.

[0056] In this way, after the current EAN receives the user equipment's ARP request in the first port and before the current EAN sends the modified ARP request through the second port after the current EAN has modified the source MAC address of the ARP request to the MAC address of the current EAN, the following may be further included: determining whether there is a punish sign for the user equipment that sent the ARP request. If yes, the ARP request will not be sent; otherwise, the current EAN modifies the source MAC address of the ARP request to the MAC address of the current EAN and sends the modified ARP request through the second port.

[0057] The example of the present disclosure shown in Fig. 4 may also support the migration of virtual machines. If it is needed to support the migration of virtual machines, then the method may further include the following:

[0058] When the current EAN receives a free ARP message from a user host in the first port, the current EAN finds corresponding ARP information according to the destination IP address of the free ARP message. If the ARP information is found, then the current EAN updates the ARP information; otherwise, the current EAN modifies the source MAC address of the free ARP message to the MAC address of the current EAN and sends the modified free ARP message through the second port.

[0059] When the current EAN receives a free ARP message from an edge device of a remote site in the second port, the current EAN finds corresponding ARP information according to the destination IP address of the free ARP message. If the ARP information is found, then the current EAN updates the ARP information and modifies the source MAC address of the free ARP message to the MAC address of the current EAN and sends the modified free ARP message through the first port. If the ARP information is not found, no processing is needed to be performed.

[0060] In the example of the present disclosure shown in Fig. 4, if a data center site is not divided into a plurality of layer 2 broadcast domains, then the user equipment is a user host in the data center.

[0061 ] If the data center site is divided into a plurality of layer 2 broadcast domains, and an edge device of each layer 2 broadcast domain is equipped with the functions of the edge device, then the user equipment is an edge device of the layer 2 broadcast domain which has the same functions as those of the edge device. [0062] The method of layer 2 interconnection between data centers of one example of the present disclosure is described above, and one example of the present disclosure also provides an edge device.

[0063] Referring to Fig. 5, there is shown a schematic diagram of an edge device according to an example of the present disclosure. The edge device includes a storage 510 and a processor 520. The storage 510 is to store computer-readable instructions including: a configuration unit 501 , a first transceiver unit 502, a second transceiver unit 503, and a processing unit 504. The processor 520 is to communicate with the storage 510 and implement the instructions stored in the storage 510.

[0064] When the processor 520 implements the instructions stored in the storage 510, the configuration unit 501 is to configure in advance and enable MFF; the first transceiver unit 502 is to: when an ARP request from a user equipment is received in a first port, send a first processing instruction to the processing unit 504, and send the ARP request through a second port after the processing unit 504 processes the ARP request according to the first processing instruction; after sending the ARP received in the first port through the second port, send a second processing instruction to the processing unit 504 if a corresponding ARP response is received in the second port, and send the ARP response through the first port after the processing unit 504 processes the ARP response according to the second processing instruction; when an ARP request from an edge device of a remote site is received in the second port, send a first processing instruction to the processing unit 504, and send the ARP request through the first port after the processing unit 504 processes the ARP request according to the first processing instruction; and after sending the ARP received in the second port through the first port, send a second processing instruction to the processing unit 504 if a corresponding ARP response is received in the first port, and send the ARP response through the second port after the processing unit 504 processes the ARP response according to the second processing instruction.

[0065] The second transceiver unit 503 is to: when a packet of which a destination MAC is the MAC address of the edge device itself from user equipment is received in the first port, send a third processing instruction to the processing unit 504, and send the packet through the second port after the processing unit 504 processes the packet according to the third processing instruction; and when a packet of which a destination MAC is the MAC address of the edge device itself from an edge device of a remote site is received in the second port, send a third processing instruction to the processing unit 504, and send the packet through the first port after the processing unit 504 processes the packet according to the third processing instruction.

[0066] The processing unit 504 is to: when receiving the first processing instruction of the first transceiver unit 502, modify a source MAC address of the ARP request received by the first transceiver unit 502 to the MAC address of the edge device itself; when receiving the second processing instruction of the first transceiver unit

502, record ARP information and modify a source MAC address of the ARP response to the MAC address of the edge device itself; when receiving the third processing instruction of the second transceiver unit

503, modify a source MAC address of the packet received by the second transceiver unit 503 to the MAC address of the edge device itself and modify a destination MAC address of the packet according to ARP information corresponding to a destination IP of the packet.

[0067] The first port is a port that is connected with user equipment within the site and the second port is a port that is connected with edge devices of remote sites.

[0068] After the processing unit 504 receives the first processing instruction of the first transceiver unit 502 and before the processing unit 504 modifies the source MAC address of the ARP request received by the first transceiver unit 502 to the MAC address of the edge device itself, the processing unit 504 is further to search corresponding ARP information in the ARP cache according to the destination IP of the ARP request. If the ARP information is found, then the processing unit 504 is to send a response instruction to the first transceiver unit 502. If the ARP information is not found, then the processing unit 504 is to modify the source MAC address of the ARP request to the MAC address of the edge device.

[0069] After the processing unit 504 processes the ARP response according to the first processing instruction and before the first transceiver unit 502 sends the user equipment's ARP request received in the first port through the second port, the first transceiver unit 502 is further to determine whether a response instruction of the processing unit 504 is received. If yes, the first transceiver unit 502 returns its own MAC address to the user equipment, otherwise, the first transceiver unit 502 sends the user equipment's ARP received in the first port through the second port. After the processing unit 504 processes the ARP response according to the first processing instruction and before the first transceiver unit 502 sends the ARP request received in the second port from the edge device of the remote site through the first port, the first transceiver unit 502 is further to determine whether the response instruction of the processing unit 504 is received. If yes, the first transceiver unit 502 returns its own MAC address to the edge device of the remote site, otherwise, the first transceiver unit 502 sends the ARP request received in the second port from the edge device of the remote site through the first port.

[0070] After the first transceiver unit 502 sends the user equipment's ARP request received in the first port through the second port, the first transceiver unit 502 is further to set a punish sign and a punish aging time for the ARP request if a corresponding ARP response is not received within a preset time and delete the punish sign for the ARP request when the aging time is exceeded.

[0071 ] After the first transceiver unit 502 receives the user equipment's ARP request in the first port and before the first transceiver unit 502 sends the first processing instruction to the processing unit 504, the first transceiver unit 502 is further to determine whether there is a punish sign for the ARP request. If yes, the first transceiver unit 502 will not send the first processing instruction to the processing unit 504, otherwise, the first transceiver unit 502 sends the first processing instruction to the processing unit 504.

[0072] After the first transceiver unit 502 sends the user equipment's ARP request received in the first port through the second port, the first transceiver unit 502 is further to set a punish sign and a punish aging time for the user equipment which sends the ARP request if a corresponding ARP response is not received within a preset time and delete the punish sign for the user equipment which sends the ARP request when the aging time is exceeded.

[0073] After the first transceiver unit 502 receives the user equipment's ARP request in the first port and before the first transceiver unit 502 sends the first processing instruction to the processing unit 504, the first transceiver unit 502 is further to determine whether there is a punish sign for the user equipment which sends the ARP request. If yes, the first transceiver unit 502 will not send the first processing instruction to the processing unit 504, otherwise, the first transceiver unit 502 sends the first processing instruction to the processing unit 504.

[0074] The first transceiver unit 502 is to send a fourth processing instruction to the processing unit 504 when the user equipment's free ARP message is received in the first port. After the processing unit 504 processes the free ARP message according to the fourth processing instruction, the processing unit 504 is to send the free ARP message through the second port if the first transceiver unit 502 receives a forwarding instruction of the processing unit 504. The first transceiver unit 502 is to send a fifth processing instruction to the processing unit 504 when a free ARP message from an edge device of a remote site is received in the second port. After the processing unit 504 processes the free ARP message according to the fifth processing instruction, the processing unit 504 is to send the free ARP message through the first port if the first transceiver unit 502 receives a forwarding instruction of the processing unit 504.

[0075] The processing unit 504 is to, after receiving the fourth processing instruction sent from the first transceiver unit 502, find corresponding ARP information according to a destination IP address of the free ARP message received by the first transceiver unit 502 in the first port, update the ARP information if the ARP information is found, otherwise, modify the source MAC address of the free ARP message to the MAC address of the edge device and send a forwarding instruction to the first transceiver unit 502. After receiving the fifth processing instruction sent from the first transceiver unit 502, find corresponding ARP information according to a destination IP address of the free ARP message received by the first transceiver unit 502 in the second port. If the ARP information is found, update the ARP information, modify the source MAC address of the free ARP message to the MAC address of the edge device and send a forwarding instruction to the first transceiver unit 502, otherwise, no processing is needed to be performed.

[0076] What has been described and illustrated herein is an example along with some of its variations. The terms, descriptions and figures used herein are set forth by way of illustration only and are not meant as limitations. Many variations are possible within the spirit and scope of the subject matter, which is intended to be defined by the following claims -- and their equivalents - in which all terms are meant in their broadest reasonable sense unless otherwise indicated.

Claims

CLAIMS: What is claimed is:

1 . A method of layer 2 interconnection between data centers, which is applied in a system comprising at least two data center networks interconnected through a backbone network, wherein each data center network accesses the backbone network through a respective edge device, and wherein each of the edge devices is enabled with media access controller (MAC) forced forwarding (MFF), the method comprising: when one of the edge devices receives an address resolution protocol (ARP) request of an edge device of a remote site in a second port, modifying, by the edge device, a source MAC address of the ARP request of the edge device of the remote site to the MAC address of the edge device; sending, by the edge device, the modified ARP request through a first port; after receiving a corresponding ARP response from the first port, recording, by the edge device, ARP information; modifying, by the edge device, a source MAC address of the ARP response to the MAC address of the edge device; and sending, by the edge device, the modified ARP response through the second port; when the edge device receives a packet of which a destination MAC is the MAC address of the edge device itself from the edge device of the remote site in the second port, modifying, by the edge device, a source MAC address of the packet to the MAC address of the edge device itself; modifying, by the edge device, the destination MAC address of the packet according to the ARP information corresponding to a destination IP of the packet; and sending, by the edge device, the modified packet through the first port; wherein the first port is a port that is connected with user equipment within a site and the second port is a port that is connected with edge devices of remote sites.

2. The method according to claim 1 , wherein: when the edge device receives an ARP request of a user equipment in the first port, modifying, by the edge device, a source MAC address of the ARP request to a MAC address of the edge device itself; and sending, by the edge device, the modified ARP request through the second port; after receiving a corresponding ARP response from the second port, recording, by the edge device, ARP information; modifying, by the edge device, a source MAC address of the ARP response to the MAC address of the edge device itself; and sending, by the edge device, the modified ARP response through the first port; when the edge device receives a packet of which a destination MAC is the MAC address of the edge device itself from the user equipment in the first port, modifying, by the edge device, a source MAC address of the packet to the MAC address of the edge device; modifying, by the edge device, a destination MAC address of the packet according to ARP information corresponding to a destination IP of the packet; and sending, by the edge device, the modified packet through the second port.

3. The method according to claim 2, wherein after the edge device receives the user equipment's ARP request in the first port and before the edge device sends the modified ARP request through the second port after the edge device has modified the source MAC address of the ARP request to the MAC address of the edge device, the method further comprises: searching corresponding ARP information in an ARP cache according to a destination IP of the ARP request; if the ARP information is found, returning the MAC address of the edge device itself to the user equipment; if the ARP information is not found, modifying the source MAC address of the ARP request to the MAC address of the edge device and sending the modified ARP request through the second port; after the edge device receives the ARP request of the edge device of the remote site in the second port and before the edge device sends the modified ARP request through the first port after the edge device has modified the source MAC address of the ARP request to the MAC address of the edge device, the method further comprises: searching corresponding ARP information in the ARP cache according to a destination IP of the ARP request of the edge device of the remote site; if the ARP information is found, returning the MAC address of the edge device to the edge device of the remote site; if the ARP information is not found, modifying the source MAC address of the ARP request to the MAC address of the edge device and sending the modified ARP request through the first port.

4. The method according to claim 3, wherein after modifying the source MAC address of the ARP request to the MAC address of the edge device and sending the modified ARP request through the second port, the method further comprises: setting a punish sign and a punish aging time for the ARP request if a corresponding ARP response is not received within a preset time, and deleting the punish sign for the ARP request when the aging time is exceeded; after the edge device receives the user equipment's ARP request in the first port and before the edge device sends the modified ARP request through the second port after the edge device has modified the source MAC address of the ARP request to the MAC address of the edge device, the method further comprises: determining whether there is a punish sign for the ARP request, in response to the punish sign for the ARP request being present, not sending the ARP request; in response to the punish sign for the ARP request not being present, modifying the source MAC address of the ARP request to the MAC address of the edge device and sending the modified ARP request through the second port.

5. The method according to claim 3, wherein after modifying the source MAC address of the ARP request to the MAC address of the edge device and sending the modified ARP request through the second port, the method further comprises: setting a punish sign and a punish aging time for the user equipment, which sends the ARP request if a corresponding ARP response is not received within a preset time, and deleting the punish sign for the user equipment which sends the ARP request when the aging time is exceeded; after receiving the user equipment's ARP request in the first port and before sending the modified ARP request through the second port after the source MAC address of the ARP request has been modified to the MAC address of the edge device, the method further comprises: determining whether there is a punish sign for the user equipment which sends the ARP request, in response to the punish sign for the ARP request being present, not sending the ARP request; in response to the punish sign for the ARP request not being present, modifying the source MAC address of the ARP request to the MAC address of the edge device and sending the modified ARP request through the second port.

6. The method according to claim 1 , wherein the method further comprises: when the edge device receives the user equipment's free ARP message in the first port, searching corresponding ARP information according to a destination IP address of the free ARP message; if the ARP information is found, updating the ARP information; if the ARP information is not found, modifying a source MAC address of the free ARP message to the MAC address of the edge device and sending the modified free ARP message through the second port; when the edge device receives a free ARP message from the edge device of the remote site in the second port, searching corresponding ARP information according to a destination IP address of the free ARP message; if the ARP information is found, updating the ARP information and modifying a source MAC address of the free ARP message to the MAC address of the edge device and sending the modified free ARP message through the first port.

7. The method according to claim 1 , wherein at least one of: the user equipment is a user host in at least one of the data centers; at least one of the data centers is divided into a plurality of layer 2 broadcast domains, and each edge device of each layer 2 broadcast domain is equipped with functions of the edge device; and the user equipment is an edge device of the layer 2 broadcast domain that has the same functions as those of the edge device.

8. An edge device applicable in layer 2 interconnection between data centers, the edge device comprising a storage and a processor, wherein the storage is to store computer-readable instructions including: a configuration unit, a first transceiver unit, a second transceiver unit, and a processing unit; the processor is to communicate with the storage and implement the instructions stored in the storage; when the processor implements the instructions stored in the storage, the configuration unit is enabled with media access controller (MAC) forced forwarding (MFF); the first transceiver unit is to, when an ARP request from an edge device of a remote site is received in a second port, send a first processing instruction to the processing unit and send the ARP request through a first port after the processing unit has processed the ARP request according to the first processing instruction; after the ARP received in the second port is sent through the first port, send a second processing instruction to the processing unit if a corresponding ARP response is received in the first port and send the ARP response through the second port after the processing unit has processed the ARP response according to the second processing instruction; the second transceiver unit is to send a third processing instruction to the processing unit when a packet of which a destination MAC is the MAC address from the edge device of the remote site is received in the second port, and send the packet through the first port after the processing unit has processed the packet according to the third processing instruction; the processing unit is to, when the first processing instruction of the first transceiver unit is received, modify a source MAC address of the ARP request received by the first transceiver unit to the MAC address of the edge device itself; when the second processing instruction of the first transceiver unit is received, record ARP information and modify a source MAC address of the ARP response to the MAC address of the edge device itself; when the third processing instruction of the second transceiver unit is received, modify a source MAC address of the packet received by the second transceiver unit to the MAC address of the edge device itself and modify a destination MAC address of the packet according to ARP information corresponding to a destination IP of the packet; wherein the first port is a port that is connected with user equipment within a site and the second port is a port that is connected with edge devices of remote sites.

9. The device according to claim 8, wherein: the first transceiver unit is further to, when an ARP request from a user equipment is received in the first port, send the first processing instruction to the processing unit and send the ARP request through the second port after the processing unit has processed the ARP request according to the first processing instruction; after the ARP received in the first port is sent through the second port, send the second processing instruction to the processing unit if a corresponding ARP response is received in the second port and send the ARP response through the first port after the processing unit has processed the ARP response according to the second processing instruction; the second transceiver unit is to send the third processing instruction to the processing unit when a packet of which a destination MAC is the MAC address of the edge device from the user equipment is received in the first port, and send the packet through the second port after the processing unit has processed the packet according to the third processing instruction.

10. The device according to claim 9, wherein after the processing unit has received the first processing instruction of the first transceiver unit and before the processing unit modifies the source MAC address of the ARP request received by the first transceiver unit to the MAC address of the edge device, the processing unit is further to: search corresponding ARP information in an ARP cache according to the destination IP of the ARP request; if the ARP information is found, send a response instruction to the first transceiver unit; if the ARP information is not found, modify the source MAC address of the ARP request to the MAC address of the edge device itself; after the processing unit processes the ARP response according to the first processing instruction and before the user equipment's ARP request received in the first port is sent through the second port, the first transceiver unit is further to determine whether the first transceiver unit has received the response instruction of the processing unit; if yes, the first transceiver unit is to return the MAC address of the edge device to the user equipment, if no, the first transceiver unit is to send the user equipment's ARP received in the first port through the second port; after the processing unit processes the ARP response according to the first processing instruction and before the ARP request received in the second port from the edge device of the remote site is sent through the first port, the first transceiver unit is further to determine whether the response instruction of the processing unit has been received; if yes, return the MAC address of the edge device itself to the edge device of the remote site, if no, send the ARP request received in the second port from the edge device of the remote site through the first port.

11 . The device according to claim 10, wherein after the first transceiver unit has sent the user equipment's ARP request received in the first port through the second port, the first transceiver unit is further to set a punish sign and a punish aging time for the ARP request if a corresponding ARP response is not received within a preset time and delete the punish sign for the ARP request when the aging time is exceeded; after the first transceiver unit has received the user equipment's ARP request in the first port and before the first transceiver unit has sent the first processing instruction to the processing unit, the first transceiver unit is further to determine whether there is a punish sign for the ARP request, if yes, the first transceiver unit is not to send the first processing instruction to the processing unit, if no, the first transceiver unit is to send the first processing instruction to the processing unit.

12. The device according to claim 10, wherein after the first transceiver unit has sent the user equipment's ARP request received in the first port through the second port, the first transceiver unit is further to set a punish sign and a punish aging time for the user equipment which sends the ARP request if a corresponding ARP response is not received within a preset time and delete the punish sign for the user equipment which sends the ARP request when the aging time is exceeded; after the first transceiver unit has received the user equipment's ARP request in the first port and before the first transceiver unit has sent the first processing instruction to the processing unit, the first transceiver unit is further to determine whether there is a punish sign for the user equipment which sends the ARP request, if yes, the first transceiver unit is not to send the first processing instruction to the processing unit, if no, the first transceiver unit is to send the first processing instruction to the processing unit.

13. The device according to claim 9, wherein the first transceiver unit is to: send a fourth processing instruction to the processing unit when the user equipment's free ARP message is received in the first port, and send the free ARP message through the second port if the first transceiver unit receives a forwarding instruction of the processing unit after the processing unit has processed the free ARP message according to the fourth processing instruction; send a fifth processing instruction to the processing unit when a free ARP message from the edge device of the remote site is received in the second port, and send the free ARP message through the first port if the first transceiver unit has received a forwarding instruction of the processing unit after the processing unit has processed the free ARP message according to the fifth processing instruction; the processing unit is to, after the fourth processing instruction sent from the first transceiver unit is received, search corresponding ARP information according to a destination IP address of the free ARP message received by the first transceiver unit in the first port, update the ARP information if the ARP information is found, otherwise, modify a source MAC address of the free ARP message to the MAC address of the edge device; and send the forwarding instruction to the first transceiver unit; after the fifth processing instruction sent from the first transceiver unit is received, search corresponding ARP information according to a destination IP address of the free ARP message received by the first transceiver unit in the second port; if the ARP information is found, update the ARP information, modify the source MAC address of the free ARP message to the MAC address of the edge device and send the forwarding instruction to the first transceiver unit.

14. A method of layer 2 interconnection between data centers which comprise at least one data center network connected with a backbone network through an edge device, wherein the edge device is enabled with media access controller (MAC) forced forwarding (MFF), and the edge device has a first port connected with user equipment within the data center network, a second port connected with the backbone network and a first MAC address; the method comprising: the edge device modifying a source MAC address of an address resolution protocol (ARP) request received in the second/first port to the first MAC address and sending the modified ARP request through the first/second port; the edge device recording ARP information of a corresponding ARP response received in the first/second port, modifying a source MAC address of the ARP response to the first MAC address and sending the modified ARP response through the second/first port; the first edge device modifying a source MAC address of a packet of which a destination MAC is the first MAC address and which is received in the second/first port to the first MAC address, modifying the destination MAC address of the packet according to the ARP information corresponding to a destination IP of the packet and sending the modified packet through the first/second port.

15. The method according to claim 14, wherein the user equipment within the data center network are indirectly connected with the backbone network through the edge device.