KR20170031539A - Apparatus and Method for Distributed Routing in Open Flow Network Partitioning Circumstance - Google Patents

Abstract

The present invention relates to a distributed routing apparatus and a method in an open flow network partitioned environment. The distributed routing method in a controller in an open flow network partitioned environment comprises: a step of managing port information of a switch managed by another controller in the same network with an external port information table; and a step of updating an accumulated cost information table for each host with accumulated cost as receiving a packet with an separately defined packet including the accumulated cost from the switch, flooding the packet to other switches, and transmitting the separately defined packet together when ports of the other switches are included in the external port information table.

Description

[0001] The present invention relates to a distributed routing apparatus and method in an open-

The present invention relates to software defined network (SDN) technology, and more particularly, to an apparatus and method capable of handling routing in a distributed manner in a distributed controller environment for high scalability and high availability.

Recently, SDN controller technology takes the form of a distributed controller that is physically composed of one or more controllers while taking the form of one central controller logically in order to support high scalability and high availability required in carrier-grade networks.

However, in order to construct a single centralized controller logically, state information of switches, links, and ports between the distributed controllers is shared. In addition, traffic and processing delay problems due to the sharing of status information between distributed controllers It is becoming an important issue that affects performance.

In other words, when the distributed controller structure is used, the performance of the network flow processing can be improved through the cooperation between the controllers. However, as the number of controllers increases, the performance degradation due to the sharing of the state information also increases, .

In the SDN distributed controller environment, when the data plane is partitioned into areas managed by respective individual controllers, the present invention provides an end-to-end routing path without sharing state information between controllers for high scalability and high availability. The present invention provides a distributed routing apparatus and method in an open flow network partitioned environment.

The present invention relates to a distributed routing method in a controller in an open flow network division environment, comprising: managing port information of a switch managed by another controller of the same network in an external port information table; The accumulated cost information table is updated with the cumulative cost, and the packet is flooded to other switches, and the ports of other switches are included in the external port information table , ≪ / RTI > sending separately defined packets together.

The present invention relates to a distributed routing apparatus in an open flow network partitioned environment, which includes an external port information table, an external port information management unit for storing port information of a switch managed by another controller of the same network in the external port information table, A cumulative cost information table, an accumulated cost information updating unit for updating a cumulative cost information table for each host with the cumulative cost, And an accumulated cost information calculation unit for generating separately defined packets when the ports of the other switches are included in the external port information table and transmitting the generated packets to the packet transmission unit.

According to the present invention, in the distributed SDN controller structure, when data planes are divided and managed by individual controllers for high scalability and high availability, state sharing messages between the distributed controllers are eliminated or minimized, And performance degradation problems can be solved.

In addition, an individual controller can set an end-to-end least cost routing path with only some partitioned topology information.

1 is a diagram showing an open-flow network division environment.
2 is a diagram illustrating an example of a process of searching for a path in an open flow network division environment.
3 is a diagram illustrating an example of a path search process in an open flow network partitioning environment according to an embodiment of the present invention.
4 is a block diagram of a controller according to an embodiment of the present invention.
5 is a flowchart illustrating an external port information management step according to an embodiment of the present invention.
6 is a flowchart for explaining a packet transmission / reception step according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the like elements throughout.

In the following description of the present invention, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear.

The terms used throughout the specification are defined in consideration of the functions in the embodiments of the present invention and can be sufficiently modified according to the intentions and customs of the user or the operator. It should be based on the contents of.

1 is a diagram showing an open-flow network division environment.

Referring to FIG. 1, one network composed of open flow switches 200-1 to 200-11 is partitioned based on proximity information of a switch. Each of the controllers 100A, 100B, and 100C manages the switches included in its own area by partitioning, searches for a link through an inter-switch link layer discovery protocol (LLDP) packet exchange, By receiving the FEATURES_REPLY packet, switch information managed by itself is learned. Therefore, each of the controllers 100A, 100B, and 100C can search for the switch and the link information managed by the controllers. In addition, the controllers 100A, 100B, and 100C can recognize the information of the host 310 when receiving the packet with the host 310 as a source for the first time.

A process of searching for a path from the source host S 310 to the destination host D 320 in the SDN controller open flow network partitioning environment will be described.

2 is a diagram illustrating an example of a process of searching for a path in an open flow network division environment.

Referring to FIG. 2, when the host S 310 transmits an ARP request message for the MAC address of the IP of the host D 320 to the host D 320, the switch 1 200 - There is no matching rate with respect to the host S 310, so the controller A 100A is inquired in the form of PACKET_IN. At this time, the controller A 100A can recognize the host S 310 and store the fact that it is connected to the switch 1 (200-1).

In addition, since the controller A 100A does not have information on the host D 320, it sets up an ARP request message for the host D 320 to float. Similarly, the controller A 100A may receive the PACKET_IN message from the switches 2, 3, 4 (200-2, 200-3, 200-4), but it already has the information about the host S 310 Do not update host information.

When the ARP request message reaches the switch 6 (200-6), the controller B 100B, like the controller A 100A, recognizes that the host S 310 is connected to the switch 6 200-6 Controller C 100C considers that host S 310 is connected to switch 8 (200-8) when first receiving an ARP message in switch 8 (200-8) in the area of controller C (100C) .

On the contrary, when the host D 320 receives the ARP request message for itself and then transmits an ARP response message in response thereto, the controller C 100C determines that the host D 320 is connected to the switch 1 200-1 . ≪ / RTI > The controller C 100C transmits a packet to the switch 8 (200-8) to which the host S 310 is connected in order to transmit the ARP response message. The controller B 100B transmits an ARP response message to the switch 8 The controller A 100A can transmit the request to the switch 1 200-1.

Finally, after the host S 310 recognizes the MAC address of the host D 320, it can transmit the data packet, and the routing path can be determined according to the location of the host D 320 stored in each of the controllers have.

Therefore, the packet transmitted from the host S 310 can be transmitted along the path of the switch 1-3-4-6-7-8-9-10.

As described above, when each of the controllers 100A, 100B, and 100C does not exchange topology state information of switches, links, ports, and the like, only the local information is used instead of the entire topology, . Therefore, although the topology information is exchanged between the controllers in the distributed controller environment, the performance degradation due to the cost due to the sharing of state information may be involved as described above.

In the present invention, an end-to-end optimal path is searched only by the local topology information of the controller without exchanging state information between the distributed controllers. ARP and DHCP messages used for host search are located in the same broadcast domain, And propagates to the whole.

FIG. 3 is a diagram illustrating a process of searching for a path in an SDN controller open flow network partitioning environment according to an embodiment of the present invention.

Referring to FIG. 3, it can be recognized that a host is connected to a switch in a path that can connect a path with the least cost between ends. That is, the minimum path (hop reference) between the host S 310 and the host D 320 is the switch 1-2-5-7-10-11, and accordingly, the controller B 100B transmits the switch 5 (200-5 The host S 310 is connected to the host D 320 and the host D 320 is connected to the switch 7 200-7. Similarly, on the basis of the controller C (100C), the host S (310) can be recognized as the switch 1 (200-1) and the host D (320)

Then, as shown in FIG. 3, a configuration of the controllers 100A, 100B, and 100C and a routing method in the controllers 100A, 100B, and 100C for transmitting a packet according to the path of the least cost among the end- .

4 is a block diagram of a controller according to an embodiment of the present invention.

4, the controller 100 includes an external port information management unit 110, an external port information table 120, a packet receiving unit 130, an accumulated cost information updating unit 140, an accumulated cost information table 150, A packet transmission unit 160 and an accumulated cost information calculation unit 170. [

The external port information management unit 110 stores port information of a switch managed by another controller of the same network in the external port information table 120. [ That is, when the switch that transmitted the LLDP packet is not a switch located in the partition managed by the external port information management unit 110 as the LLDP packet is received, the switch that transmitted the LLDP packet is managed by another controller in the same network If the switch is a switch managed by the distributed controller in the same network, the port information of the switch is stored in the external port information table. Here, in order to determine whether the switch is managed by another controller in the same network, the external port information management unit 110 determines whether the ID of the controller, open flow support, or the like, through the Optional TLV field defined in the LDP packet, .

Here, the external port information table 120 may be configured as shown in Table 1 below.

No DPID of Switch Port #

The packet receiving unit 130 receives a separately defined packet including an accumulated cost together with the packet from the switch. In this case, when the packet received from the switch includes a separately defined packet including the accumulated cost information from the sender to the switch, the packet receiving unit 130 separates the separately defined packet and updates the accumulated cost information updating unit 140, .

The accumulated cost information updating unit 140 updates the accumulated cost information table 150 for each host with the accumulated cost. At this time, the accumulated cost information updating unit 140 compares the cumulative cost information for each host stored in advance, and updates the cumulative cost information table for each host when the cost information is minimum for the same host.

The cumulative cost information table 150 stores the cost from the sender of the L2 packet to the current switch as shown in Table 2 below. The cumulative cost is the cumulative cost corresponding to the sender's entry corresponding to the Host IP address. switch value and the path cost from the corresponding switch (DPID) to the current switch.

No Host IP address corresponding switch's DPID Cumulative cost to corresponding switch

The packet transmission unit 160 floods the packet to other switches. That is, the packet transmission unit 150 searches whether the ports of other switches are included in the external port information table 120, and when the port is included in the external port information table 120, To acquire the accumulated cost information, and transmits the separately defined packet including the acquired accumulated cost information together with the packet. If the port of the switch corresponding to the search result is not included in the external port information table 120, the packet transmission unit 160 transmits only the packet to the corresponding port.

The cumulative cost information calculation unit 170 calculates the cumulative cost at the request of the packet transmission unit 160 and provides the cumulative cost to the packet transmission unit 160. The cumulative cost is calculated as the sum of the cumulative cost to corresponding switch value of the sender entry corresponding to the Host IP address and the path cost from the corresponding switch's DPID to the current switch as shown in Table 2 do.

Next, a distributed routing method in an open flow network partitioned environment can be largely comprised of an external port information management step and a packet transmission / reception processing step.

5 is a flowchart illustrating an external port information management step according to an embodiment of the present invention.

Referring to FIG. 5, when the controller receives the LLDP packet (S510), it determines whether the transmission switch of the LLDP packet is transmitted by a switch located in the partition managed by the controller (S520).

If it is determined in step S520 that the LLDP packet is not the LLDP packet transmitted by the switch in the partition managed by itself, the controller determines whether the LLDP packet is transmitted by a switch managed by another controller in the same network (S530). Since the Optional TLV field is defined in the LLDP packet, it can be determined whether the controller ID, OpenFlow support, and whether it is a distributed controller that manages the same network.

If it is determined in step S530 that the switch is the LLDP transmitted by the switch managed by the distributed controller in the same network, the controller stores the port information of the switch in the external port information table (S540). Here, the external port information table includes the DPID of the switch and the port number.

6 is a flowchart for explaining a packet transmission / reception step according to an embodiment of the present invention.

Referring to FIG. 6, the controller receives the L2 packet from the switch (S610), and determines whether the accumulated cost packet is a separately defined packet (S620). Here, when the switch receives a packet defined separately from a switch of a neighboring partition, since there is no corresponding matching rule, the controller can query the controller in the form of PACKET_IN, so that the controller can determine whether or not it is a packet defined separately .

If it is determined in operation S620 that the L2 packet is a separately defined packet that is an accumulated cost packet, the controller updates the accumulated cost information table for each host by separating the L2 packet and the cost information (S640). Also, it stores information that the discovered host information is connected to the corresponding switch, like the general controller procedure.

At this time, since the controller can receive packets defined separately from other switches in its own partition, if the cost information is small for the same host (S630), the cumulative cost information table and the host information for each host are updated .

In step S660, the controller searches the external port information table (S650) before the L2 packet, such as an ARP request, is flooded (transmitted to all ports of the switch), and determines whether the port is included in the external port information table.

If it is determined in step S660 that the corresponding switch port is included in the external port information table, the controller transmits a separately defined packet to the corresponding port so that the controller can interpret the switch port (S680). In other words, together with the L2 packet, cost information from the sender of the L2 packet to the current switch is transmitted.

On the other hand, if it is determined in step S660 that the port of the corresponding switch is not included in the external port information table, the controller directly transmits the L2 packet to the port (S670).

At this time, the cost from the sender of the L2 packet to the current switch is calculated as the sum of the path cost from the corresponding switch's corresponding DPID to the cumulative cost to corresponding switch value of the sender entry corresponding to the host IP address . If the sender of the L2 packet does not exist in the cumulative cost information table for each host, it means that the host is connected to the switch in the partition managed by the individual controller. have.

Instead of transmitting a packet defined separately through the above-described routing method to the data plane, a separately defined packet is transmitted through sharing of state information between the controllers, and the L2 packet received by the controller receiving the packet is flooded At the same time, the following procedure can be performed. In this case, the inter-controller state information sharing mechanism can be applied.

The above-described method assumes that one network composed of open flow switches is partitioned based on the proximity information of the switches. To do so, one of the individual controllers may allow the connection of the switches on the data plane and migrate the switches on the data plane to the individual controllers in the distributed controller in real time or after a certain time. In addition, a device may be added to collect and control information of individual controllers to allow the user to monitor and control the entire topology through the Northbound API (API).

Claims (17)

A distributed routing method in a controller in an open flow network partitioned environment,
Managing port information of a switch managed by another controller of the same network in an external port information table;
And a controller for updating the cumulative cost information table for each host with the cumulative cost and for flooding the packet to other switches, And if the packet is included in the external port information table, transmitting separately defined packets together.
The method of claim 1, wherein the administering comprises:
Determining whether the switch transmitting the LLDP packet is a switch located in a partition managed by the LLDP packet upon receiving the LLDP packet,
Determining whether a switch that has transmitted the LLDP packet is a switch managed by another controller in the same network, if it is determined that the switch is not a switch in a partition managed by itself;
And storing the port information of the switch in an external port information table when the switch is a switch managed by a distributed controller in the same network as the determination result.
The method as claimed in claim 2, wherein the step of determining whether the switch is managed by another controller in the same network includes: determining whether the ID of the controller, open flow support, and whether the distributed controller manages the same network through the Optional TLV field defined in the LLDP packet Wherein said method comprises the steps of:
The method of claim 2, wherein the external port information table
And a DPID and a port number of the switch. The distributed routing method in an open-flow network division environment.
2. The method of claim 1, wherein the transmitting
Determining whether a packet received from the switch includes a separately defined packet including cumulative cost information from the sender to the switch;
Updating a cumulative cost information table for each host by separating a packet defined separately from the packet when the separately defined packet is included;
The method comprising the steps of: flooding the packet to other switches, and detecting whether a port of another switch is included in the external port information table;
And if the search result port is included in the external port information table, transmitting a packet separately defined together with the packet to the corresponding port.
6. The method of claim 5, wherein the updating comprises:
Wherein the accumulated cost information for each host is compared with the accumulated cost information for each host, and the accumulated cost information table for each host is updated when the cost information is minimum for the same host.
6. The method of claim 5,
Further comprising the step of transmitting only a packet to the corresponding port when the port of the corresponding switch is not included in the external port information table as a result of the search.
6. The method of claim 5,
And the path cost from the corresponding switch (corresponding switch's DPID) to the current switch is calculated as the cumulative cost to corresponding switch value of the sender entry corresponding to the host IP address. Way.
An external port information table,
An external port information management unit for storing port information of a switch managed by another controller of the same network in the external port information table;
A packet receiver for receiving a separately defined packet including an accumulated cost together with a packet from the switch;
An accumulated cost information table,
An accumulated cost information updating unit for updating the accumulated cost information table for each host by the accumulated cost;
A packet transmitter for flooding the packet to other switches;
And an accumulated cost information calculation unit for generating a separately defined packet and transmitting the separately defined packet to the packet transmission unit when a port of another switch is included in the external port information table. Device. The method of claim 9, wherein the external port information management unit
When the LLDP packet is received, if the switch transmitting the LLDP packet is not a switch located in the partition managed by the LLDP packet, it is determined whether the switch that transmitted the LLDP packet is a switch managed by another controller in the same network. And the port information of the switch is stored in the external port information table when the switch is managed by the internal distributed controller.
The method of claim 10, wherein the external port information management unit
In order to determine whether the switch is managed by another controller in the same network, it is determined whether the ID of the controller, the open flow support, and whether it is a distributed controller managing the same network through the Optional TLV field defined in the LDP packet Distributed routing apparatus in a flow network partitioned environment.
11. The method of claim 10, wherein the external port information table
And a DPID and a port number of the switch. The apparatus of claim 9, wherein the packet receiver
When the packet received from the switch includes a separately defined packet including cumulative cost information from the sender to the switch, separates the separately defined packet and delivers it to the cumulative cost information updating unit. Distributed routing device in an environment.
14. The method of claim 13, wherein the cumulative cost information updating unit
Wherein the accumulated cost information for each host is compared with the accumulated cost information for each host, and the accumulated cost information for each host is updated when the cost information is the minimum for the same host.
The method of claim 9, wherein the packet transmission unit
The packet is flooded to other switches, and a check is made as to whether ports of other switches are included in the external port information table. When the port is included in the external port information table, Cost information, and transmits a separately defined packet including the acquired accumulated cost information together with the packet.
16. The apparatus of claim 15, wherein the packet transmitter
Wherein when the port of the corresponding switch is not included in the external port information table, only the packet is transmitted to the corresponding port.
9. The method of claim 8,
And the path cost from the corresponding switch (corresponding switch's DPID) to the current switch is calculated as the cumulative cost to corresponding switch value of the sender entry corresponding to the host IP address. Device.

Images