CN103326943B - Data flow heavy route method and controller - Google Patents
Data flow heavy route method and controller Download PDFInfo
- Publication number
- CN103326943B CN103326943B CN201210080081.1A CN201210080081A CN103326943B CN 103326943 B CN103326943 B CN 103326943B CN 201210080081 A CN201210080081 A CN 201210080081A CN 103326943 B CN103326943 B CN 103326943B
- Authority
- CN
- China
- Prior art keywords
- switch
- location address
- address
- data flow
- main frame
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The present invention proposes a kind of data flow heavy route method, including：According to position in a network, it is switch and host assignment location address, different location address represents different routed paths；Select the location address of a routed path and respective host for each data flow, the location address of selected routed path and respective host is stored in stream information table, the address translation rule between the physical address of main frame and the location address of main frame is stored in the flow table of the edge switch related with main frame；And when network topology changes, find out affected all data flows, select the location address of new routed path and respective host for each data flow, routed path in renewal stream information table and the location address of respective host, and the address translation rule in the flow table of the edge switch updating this data flow.The invention allows for a kind of controller for realizing above-mentioned data flow heavy route method.
Description
Technical field
The present invention relates to data network, in particular it relates to a kind of data flow heavy route method and controller.
Background technology
In traditional switch or router, quick packet forwards (data path) and high-rise routing decision
(control path) is completed by same equipment.Because it is next independent complete that switch or router need to have enough functions
Realizing of one-tenth operations, the therefore network equipment is extremely complex.In addition, network vendors typically will not open switch or route
The source code of device.When third party's exploitation or when increasing new function, the closure of system is to the idea of novelty and new technique
Enter and cause very high threshold.
In order to solve this problem it is proposed that a kind of new network architecture (referred to as OpenFlow).In OpenFlow network
In, the data path of switch and control path are separated.Datapath section still on switches, and determine by high-level route
Plan is then completed by a single controller.Pass through OpenFlow protocol communication between OpenFlow switch and controller.
OpenFlow provides a kind of general mode for research worker and carrys out running experiment, verifies their new idea, and without manufacturer
The internal work mechanism of the open network equipment.
The OpenFlow network architecture comprises OpenFlow switch and controller, passes through OpenFlow agreement between the two and leads to
Letter.OpenFlow switch comprises a network data flow table (flow table) and an escape way (secure
channel).OpenFlow switch carries out forwarding and the process of packet by the content in Network Search data stream list,
Communicated with controller by escape way.Flow table contains a series of flow table item (flow entry), and flow table item is generally speaking
Including three partial contents：1) packet header field, for matched data bag, including inbound port, source MAC, target MAC (Media Access Control) address,
The items such as source IP address, purpose IP address；2) enumerator；And 3) action-item (action), for indicating that packet matches it
Operation afterwards.
Fig. 1 shows the example of a network topology 100, comprises switch S1～S4 and main frame H1 in network topology 100
And H2, controller 120 and switch S1～S4 communication.In this network topology, main frame H1 and H2 communicates.
Fig. 2 shows packet repeating process during main frame H1 and H2 communication in the network topology 100 shown in Fig. 1.System
When system starts, the flow table of switch S1～S4 is all empty, and switch S1 first receives the packet (step that main frame H1 sends
S201), search the flow table of oneself to be mated, now flow table is sky, does not match (step S202), this is counted by switch S1
It is sent to controller 120 (step S203) according to bag.Controller 120 be this packet calculate route, have selected H1 → S1 → S2 →
This paths of S3 → H2 (step S204), then sends commands to switch S1, S2 and S3, allows them update flow table (step
S205).After receiving instruction, switch S1, S2 and S3 add flow table item (step S206) accordingly, and the flow table item newly increasing is used
Data flow when main frame H1 and H2 is communicated is processed.In step S207, this packet is returned to friendship by controller 120
Change planes S1, and indicates that it forwards from port p1.This packet is sent to S2 (step S208) according to instruction by switch S1.Exchange
After machine S2 receives packet, search the flow table of oneself and mated, match the flow table item (step S209) of new interpolation, will
Packet is transmitted to switch S3 (step S210).Switch S3 searches the flow table of oneself, has also matched the flow table of new interpolation
Item (step S211), packet is transmitted to main frame H2 (step S212).This is the processing procedure of first packet, for this
The follow-up packet of data flow, switch S1, S2 and S3 can match corresponding flow table item.
In network, data flow heavy-route is a very common demand.Such as, in the example network topology 100 shown in Fig. 1,
Main frame H1 and H2 communicates, and the communication path that controller 120 replaces them to select is H1 → S1 → S2 → S3 → H2.If this paths
On certain link break down, in order to ensure the proper communication of main frame H1 and H2, be necessary for the data stream gravity between them
Be routed to one new up to path on.
Fig. 3 A and Fig. 3 B shows the example of data flow heavy-route.For example, as shown in Figure 3A, the network shown in Fig. 1 is opened up
Flutter in 100, the existing route of main frame H1 and H2 communication is H1 → S1 → S2 → S3 → H2 (before heavy-route).If switch S1 with
Link failure between S2, as shown in Figure 3 B, this stream gravity can be routed on path H1 → S1 → S4 → S3 → H2 for system
(after heavy-route), Fig. 4 shows the sequential chart of this data flow heavy-route process.Before also show heavy-route in Fig. 3 A and Fig. 3 B
Afterwards, the content changing in the flow table of switch S1～S4.
With reference to Fig. 4, controller 120 is operated by periodic Topology Discovery, finds link failure or exchange fault (step
Rapid S401).Controller 120 search stream information table, find out all streams (step S402) being affected by this fault, afterwards, be by
Each stream of impact selects new path, updates stream information table, the new path (step S403) selecting of storage.For Mei Tiaochong road
By stream, all switches on the existing route that controller 120 flows to this and the new path selecting send instruction, and instruction updates
Flow table (step S404).The all switch S1～S4 receiving instruction from controller 120, according to this instruction, increase to flow table
New flow table item, or delete existing flow table item (step S405).Hereafter, for the follow-up data bag of the stream being by dropping rerouted,
Switch S1～S4 forwards packet (step S406～S412) according to new flow table item.
During data flow heavy-route shown in Fig. 3 A, Fig. 3 B and Fig. 4, controller 120 needs on routed path
All switch S1～S4 sends instruction, indicates that it updates flow table.In the case that network size and dynamic are all very big, so
Reroute operation all can bring very big burden to controller and network.
Content of the invention
The present invention proposes a kind of data flow heavy route method.By the applicant in Chinese invention patent application number CN
The location address proposing in 201110077135.4 or other addressing mechanism with similar feature are used for network addressing, position
Address replaces MAC Address to be used for packet forwarding in a network, and edge switch is responsible for MAC Address and position of host machine address
Conversion.Address translation rule when system needs to carry out heavy-route to a data stream it is only necessary on modification edge switch.
Such as, old address translation rule on edge switch is：" target MAC (Media Access Control) address [destMAC] is revised as the position of main frame H1
Address 1 [loc1 (H1)] ", after modification, new transformational rule is：" target MAC (Media Access Control) address [destMAC] is revised as the position of main frame H1
Put address 2 [loc2 (H1)] ".Because diverse location address loc1 (H1) as the main frame H1 on packet rs destination ground and loc2
(H1) different routed paths are represented respectively, therefore in this way it is possible to realize the data between main frame H1 and H2
Stream gravity route.
First scheme according to the present invention it is proposed that a kind of data flow heavy route method, including：According to position in a network
Put, be switch and host assignment location address, different location address represents different routed paths；For the choosing of each data flow
Select the location address of a routed path and respective host, by the location address storage of selected routed path and respective host
In stream information table, the address translation rule between the physical address of main frame and the location address of main frame is stored in and main frame phase
In the flow table of edge switch closed；And when network topology changes, find out affected all data flows, for every
Individual data flow selects the location address of new routed path and respective host, updates the routed path in stream information table and accordingly leads
Address translation rule in the location address of machine, and the flow table of the edge switch updating this data flow.
Preferably, network topology changes including determining newly-increased switch by Topology Discovery, when newly-increased switch
When, the location address extra to newly-increased switch and the switch related to newly-increased switch and host assignment, more new position ground
Location table；And find out all data flows needing heavy-route according to the requirement of load balancing, select new road for each data flow
Location address by path and respective host.
Preferably, at least one main frame has two location address.
Preferably, described data flow heavy route method, also includes：Based on the location address being distributed, it is the institute in network
There is switch to calculate the flow table forwarding for packet, and the flow table calculating is installed in switch.
Preferably, network topology is changed, including being determined by Topology Discovery, link failure or switch event occurs
Barrier.
Alternative plan according to the present invention it is proposed that a kind of controller, including：Switch management unit, exists for basis
Position in network, is switch and host assignment location address, produces location address table, and different location address represents difference
Routed path, described switch management unit is additionally operable to select the position of a routed path and respective host for each data flow
Put address, the location address of selected routed path and respective host is stored in stream information table, and the physics by main frame
Address translation rule between address and the location address of main frame notifies to give the related edge switch of main frame；Topology Discovery list
Unit, for finding that protocol data bag finds network topology by indicating all switch transmission link layers, produces annexation
Table；Memorizer, for storing connection relational table, location address table and stream information table；And heavy-route unit, for by checking
Connection relational table finds network topological change, finds out affected all data flows, selects new route for each data flow
Path and the location address of respective host, the routed path in renewal stream information table and the location address of respective host, and will be right
The update notification of address translation rule gives the edge switch of this data flow.
Preferably, described switch management unit is additionally operable to when by checking that connection relational table finds newly-increased switch,
The location address extra to newly-increased switch and the switch related to newly-increased switch and host assignment, updates location address
Table, and described heavy-route unit is additionally operable to, in newly-increased switch, find out and need heavy-route according to the requirement of load balancing
All data flows, select the location address of new routed path and respective host for each data flow.
Preferably, described switch management unit is that at least one main frame addresses two location address.
Preferably, described switch management unit is additionally operable to based on the location address being distributed, and is all friendships in network
Change planes and calculate the flow table forwarding for packet, and the flow table calculating is installed in switch.
Preferably, network topological change includes determining generation link failure or switch event by checking connection relational table
Barrier.
In existing data flow heavy route method, controller needs to indicate that all switches on routed path update stream
Table.And it is only necessary to the address on two edge switch of modification turns in new data flow heavy route method proposed by the present invention
Change rule, without the flow table of other switches updating on routed path.
Brief description
By the preferred embodiments of the present invention are described below in conjunction with the accompanying drawings, above and other purpose, the spy of the present invention will be made
Advantage of seeking peace is clearer, wherein：
Fig. 1 shows the example of a network topology 100.
Fig. 2 shows packet repeating process during main frame H1 and H2 communication in the network topology 100 shown in Fig. 1.
Fig. 3 A and Fig. 3 B shows the example of data flow heavy-route, before and after wherein also show heavy-route, switch S1～
Content changing in the flow table of S4.
Fig. 4 shows the sequential chart of data flow heavy-route process.
Fig. 5 shows the example of location address.
Fig. 6 shows the schematic construction block diagram of switch 8100 according to the present invention and controller 8200.
Fig. 7 shows the example of the flow table set up on switch S2 and S4.
Fig. 8 A and Fig. 8 B shows the example of the data flow heavy-route according to the present invention, before wherein also show heavy-route
Afterwards, the content changing in the flow table of switch S1～S4.
Fig. 9 shows the sequential chart of the data flow heavy-route process according to the present invention.
Figure 10 shows the schematic construction block diagram of switch C100 according to the present invention and controller C200.
Figure 11 shows the method flow diagram carrying out data flow heavy-route due to the network capacity extension.
Figure 12～Figure 15 shows the example of the network capacity extension.
In all the drawings in the present invention, same or analogous structure and step are all with same or analogous reference mark
Know.
Specific embodiment
With reference to the accompanying drawings to a preferred embodiment of the present invention will be described in detail, eliminate in the course of the description for this
It is unnecessary details and function for invention, to prevent the understanding of the present invention from causing to obscure.
Another Chinese invention patent application (application number CN 201110077135.4) of the applicant proposes one kind and is used for
The addressing method of data center network.Main frame in network and switch are allocated one or more location address
(locator), location address is stratification, encodes the main frame and switch position in network topology.Work as network topology
Provide redundancy path when, distribute multiple location address to main frame and switch.Location address replaces MAC Address to be stored in friendship
In the flow table changed planes, for the forwarding of packet.Multiple location address describe mulitpath, and it is suitable to select in mulitpath
Path can achieve load balancing.
Fig. 5 shows the example of location address, and network topology 500 is identical with the network topology 100 shown in Fig. 1.Main frame H1
There are two paths and H2 between, therefore main frame H1 and H2 has also been respectively allocated two location address.
Fig. 6 shows the schematic construction block diagram of switch 6100 according to the present invention and controller 6200.
As shown in fig. 6, switch 6100 includes：Local management unit 6110, memorizer 6120 data packet forwarding unit
6130.
Local management unit 6110 is responsible for being communicated with controller 6200, and the instruction being sent according to controller 6200
Update the flow table 612 being stored in memorizer 6120.When needed (for example, during system start-up, when no mating list item, etc.),
Local management unit 6110 can deliver a packet to controller 6200.In addition, according to the instruction of controller 6200, local pipe
Packet is supplied to data packet forwarding unit 6130 by reason unit 6110.
Memorizer 6120 storage flow table 612.According to the present invention, the flow table 612 of non-edge switch only comprises flow table item；And
The flow table 612 of edge switch not only comprises flow table item, also comprises address translation rule list item, according to address translation rule list item
The address translation rule being stated clearly, realizes the conversion between MAC Address and location address.
Data packet forwarding unit 6130 according to the content in flow table 612 or the instruction according to local management unit 6110,
Send data packets to corresponding output port.
As shown in fig. 6, controller 6200 includes：Switch management unit 6210, Topology Discovery unit 6220, memorizer
6230 and heavy-route unit 6240.
Switch management unit 6210 is responsible for communicating with switch 6100, and realizes the management to switch 6100.For example,
Management to switch 6100 includes but is not limited to：According to the topological structure of network, it is the switch 6100 in network and correlation
Host assignment location address, and store these location address；Based on the location address being distributed, it is all switches in network
6100 calculating flow tables, and the flow table calculating is installed in switch 6100；Select a routed path for each data flow
With the location address of respective host, the location address of selected routed path and respective host is stored in stream information table 6236
In；The address translation rule and location address of main frame between for the physical address (such as MAC Address) of main frame is notified to give master
The related edge switch of machine.
Topology Discovery unit 6220 passes through to indicate that all switch 6100 transmission link layers find agreement (LLDP) packet
To find network topology.For example, when a certain switch 6100 is added in network, this switch 6100 can active to-controller
6200 set up connection.Controller 6200 periodicity indicates that all switches 6100 send LLDP data from their all of the port
Bag.After each switch 6100 receives LLDP packet, this packet is sent to controller 6200.Controller 6200 by
This obtains the topology of whole network, can be according to [(switch 1, the port of switch 1, switch 2, the end of switch 2
Mouthful) ...] form connection relational table is stored in memorizer 6230 (network topology 6232).For example, the net shown in Fig. 5
In network example topology, the network topology 6232 (connection relational table) obtaining is：[(S1, p1, S2, p1), (S1, p2, S4, p2),
(S2, p2, S3, p3), (S3, p2, S4, p1)].
Memorizer 6230 storage network topology 6232 (connection relational table), location address table 6234 and stream information table 6236.
Network topology 6232 stores the topology information of network, has stated clearly the annexation between all switches.Location address table
6234 location address storing switch 6100 and main frame in network, location address encodes the positional information of network node,
And it is used for packet forwarding in a network.Stream information table 6236 store network in survival data flow information, including for
Every path flowing selection, and include for every location address flowing selection.
Heavy-route unit 6240 is directed to the change of network topology 6232, and data flow is re-routed to new road from original path
Footpath.According to the present invention, in heavy-route, heavy-route unit 6240 only needs to change the address translation rule list item on edge switch
The address translation rule being stated clearly.By heavy-route, it is possible to achieve the purpose of network fault tolerance.Heavy-route unit 6240 passes through to check
Network topology 6232 (connection relational table) finds network topological change, finds out affected all data flows, for each data
Stream selects the location address of new routed path and respective host, updates the routed path in stream information table 6236 and respective host
Location address, and the update notification of address translation rule will be given with the edge switch of this data flow.
Before the practical communication between main frame starts, system needs an initialization procedure.First, in controller 6200
Topology Discovery unit 6220 Topology Discovery is carried out to network, obtain the topology of whole network.Then, the friendship in controller 6200
Administrative unit of changing planes 6210, according to the topological structure of network, is the switch 6100 in network and host assignment location address, and
Location address is stored in position address table 6234.Then, the switch management unit 6210 in controller 6200 is network
In all switches 6100 calculate flow table, these flow tables are based on location address, and form is：Destination locations address → output
Port.For example, Fig. 7 shows the example of the flow table set up on switch S2 and S4.Switch management in controller 6200
The flow table calculating is installed in the flow table of switch 6100 unit 6210, completes initialization procedure.Follow-up packet is
Forwarded in a network based on location address, rather than traditional MAC Address, on edge switch, the address translation rule of storage can
To realize the conversion between MAC Address and location address.Switch management unit 6210 in controller 6200 is also responsible for often
Individual data flow selects the location address of a routed path and respective host, by the position of selected routed path and respective host
Put address to be stored in stream information table 6236.
Fig. 8 A and Fig. 8 B shows the example of the data flow heavy-route according to the present invention.For example, as shown in Figure 8 A, in Fig. 5
In shown network topology 500, the existing route of main frame H1 and H2 communication is H1 → S1 → S2 → S3 → H2 (before heavy-route).False
As the link failure between switch S1 and S2, this stream gravity can be routed on path H1 → S1 → S4 → S3 → H2 for system
(after heavy-route), Fig. 9 shows the sequential chart of this data flow heavy-route process.Before also show heavy-route in Fig. 8 A and Fig. 8 B
Afterwards, the content changing in the flow table of switch S1～S4.
With reference to Fig. 9, Topology Discovery unit 6220 is operated by periodic Topology Discovery, updates network topology 6232.Weight
Routing unit 6240 passes through to check network topology 6232, finds link failure or exchange fault (step S901).Heavy-route list
Unit 6240 lookup stream information table 6236, finds out all streams (step S902) being affected by this fault, afterwards, heavy-route unit
6240 is each stream affected, selects the new location address of new routed path and respective host, updates stream information table
6236, path and new location address (step S903) that storage newly selects.For the stream of every heavy-route, switch management list
Unit 6210 only needs edge switch S1 and S3 (access switch and egress switch) that this is flowed to send instruction, and instruction edge is handed over
S1 and S3 that change planes updates flow table 612 (step S904).Receive the edge switch S1 of instruction and the S3 root from controller 6200
According to this instruction, change the address translation rule list item (step S905) in flow table.Hereafter, for the follow-up number of the stream being by dropping rerouted
According to bag, edge switch S1 and S3 according to new address translation rule, realize source MAC and amended source position address it
Between conversion, and realize the conversion between target MAC (Media Access Control) address and amended destination locations address, and according to new position ground
Flow table item is searched in location, forwards packet (step S906～S908 and S911～S912).Non-edge switch S2 and S4 is according to new
Location address searches flow table item, forwards packet (step S909～S910).
By Fig. 8 B with Fig. 3 B shown in situation compared with, address translation rule table in only two edge switch S1 and S3
Item there occurs change.
Data flow heavy route method proposed by the present invention can be additionally used in the on-line rapid estimation of the network architecture, and system self-adaption is in new
Network topology, automatically carry out the load balancing of data flow, without causing the interruption of top service.
Figure 10 shows the schematic construction block diagram of switch A 100 according to the present invention and controller A200, for realizing
The on-line rapid estimation of the network architecture.In Fig. 10, for purposes of brevity, unit same as shown in Figure 6 identical accompanying drawing mark
Note represents, and eliminates the detailed description to these units.
As shown in Figure 10, switch A 100 includes：Local management unit 6110, memorizer 6120, data packet forwarding unit
6130 and link information collector unit A140.
Link information collector unit A140 collects the link-state information of all links of this switch A 100, including every
The flow information of link, is then reported to controller A200 by local management unit 6110, and link information is collected and reported can
To be realized using the standard feature defined in OpenFlow specification and agreement.
As shown in Figure 10, controller A200 includes：Switch management unit A210, Topology Discovery unit 6220, memorizer
A230 and heavy-route unit A240.
Switch management unit A210, in addition to having the function of switch management unit 6210, also has for new exchange
Machine and the switch related to newly-increased switch and the function of the extra location address of host assignment, these location address are to existing
The increment addressing of the location address of some network nodes, increment addressing refers to not affect location address and the stream of existing network node
Table it is only necessary to add new location address and flow table item, without modification or delete existing location address and flow table item.
Memorizer A230 storage network topology 6232 (connection relational table), location address table 6234, stream information table 6236 and
Link information table A236.Link information table A236 stores the flow information of each of the links in network.
Heavy-route unit A240 in addition to there is the function of heavy-route unit 6240 additionally it is possible to when carrying out the network capacity extension,
Based on the link information being stored in link information table A236, some data flows are re-routed to load from the path of load weight light
Path on.
Figure 11 shows the method flow diagram carrying out data flow heavy-route due to the network capacity extension.
When due to some of network congestion locally occurring, and when congestion local increases switch, Topology Discovery unit
6220 are operated by periodic Topology Discovery, update network topology 6232.Heavy-route unit A240 passes through to check network topology
6232, find the switch (step S1101) newly increasing.Switch management unit A210 give new switch and with newly-increased exchange
The related switch of machine and the extra location address of host assignment, these location address are the position ground to existing network node
The increment addressing of location, updates location address table 6234 (step S1102).Then, switch management unit 6210 calculates new adding
The flow table of switch, and the flow table item that associated switch needs newly increase, and these new flow table items are notified to corresponding
Switch (step S1103).Heavy-route unit A240, according to the information in stream information table 6236 and link information table A236, looks for
Go out to need all data flows (step S1104) of heavy-route according to the requirement of load balancing.Heavy-route unit A240 is every need
Want the stream of heavy-route, select the new location address of new routed path and respective host, update stream information table 6236, storage is new
The path selecting and new location address (step S1105).For the stream of every heavy-route, switch management unit 6210 only needs
The edge switch (access switch and egress switch) that this is flowed sends instruction, and instruction edge switch updates flow table 612
(step S1106).
The switch receiving the newly-increased list item instruction from controller 6200 instructs according to this newly-increased list item, to flow table 612
The new flow table item (step S1107) of middle increase.The edge switch receiving the modification instruction from controller 6200 is repaiied according to this
Change instruction, the address translation rule list item (step S1108) in modification flow table.Hereafter, for the follow-up number of the stream being by dropping rerouted
According to bag, edge switch, according to new address translation rule, realizes turning between source MAC and amended source position address
Change, and realize the conversion between target MAC (Media Access Control) address and amended destination locations address, and search according to new location address
Flow table item, forwards packet；Non-edge switch, according to new position address search flow table item, forwards packet (step S1109).
Figure 12 shows the example of the network capacity extension.In legacy network topological structure, 1 core switch is connected to 3
Edge switch, and then it is connected to 6 main frame H1～H6.As shown in figure 12,4 data stream F1～F4 is through same link,
Cause congestion (dotted-line ellipse frame).A core switch can be increased newly, heavy-route is carried out to data flow F2 and F4, alleviation is gathered around
Plug.It is possible to further increase an edge of table switch newly again, heavy-route is carried out to data flow F3, alleviate congestion further.
Figure 13, Figure 14, Figure 15 sets forth these three stages, and (legacy network topological structure, a newly-increased core exchange
Machine, increase an edge of table switch again newly), the flow table item in the location address of network node, switch flow table and edge exchange
The content changing of the address translation rule list item on machine.
Taking the flow table of edge switch S3 as a example it is considered to data flow (the flow table of S3 from source host H4 to destination host H1
In the third line).
In legacy network topological structure (Figure 13), the MAC Address (source MAC) of source host H4 is revised as source host H4
Location address 1 (loc1 (H4)), by the MAC Address (purpose MAC) of destination host H1 be revised as destination host H1 position ground
Location 1 (loc1 (H1)), afterwards, from port p3 output.
After a newly-increased core switch S5 (Figure 14), source host H4 and destination host H1 is all assigned extra
Location address, i.e. location address 2.Now, the MAC Address (source MAC) of source host H4 is revised as the location address 2 of source host H4
(loc2 (H4)), the MAC Address (purpose MAC) of destination host H1 is revised as the location address 2 (loc2 of destination host H1
(H1)), afterwards, export from port p4.Thus, this data flow is routed to newly-increased core switch S5.
After increasing by edge of table switch S6 (Figure 15) further, source host H4 still only has two location address
(identical with Figure 14), but destination host H1 is assigned with two extra location address again, i.e. location address 3 and location address
4.Now, the MAC Address (source MAC) of source host H4 is revised as the location address 2 (loc2 (H4)) of source host H4, by purpose master
The MAC Address (purpose MAC) of machine H1 is revised as the location address 4 (loc4 (H1)) of destination host H1, and afterwards, from port, p4 is defeated
Go out.Thus, this data flow is routed to newly-increased core switch S5, and and then is routed to newly-increased edge switch S6.
For the flow table of other switches, similar description can be obtained, for brevity, will not be described here.
Other settings of the embodiment of the present invention disclosed herein include executing the method formerly summarized and subsequently describe in detail in fact
Apply the software program of the steps and operations of example.More specifically, computer program is a kind of following embodiment：There is calculating
Machine computer-readable recording medium, on computer-readable medium, coding has computer program logic, when executing on the computing device, computer journey
Sequence logic provides related operation, thus providing above-mentioned data flow rerouting scheme.When at least one process in computing system
When executing on device, computer program logic makes the operation (method) described in the computing device embodiment of the present invention.The present invention's
The computer that this set is typically provided as arranging or encoding in such as optical medium (such as CD-ROM), floppy disk or hard disk etc. can
On software on reading medium, code and/or other data structures or such as one or more ROM or RAM or PROM chip
Downloadable software in other media of firmware or microcode or special IC (ASIC) or one or more module
Image, shared data bank etc..Software or firmware or this configuration may be installed in computing device, so that in computing device
Technology described in the individual or multiple computing device embodiment of the present invention.In conjunction with such as one group of data communications equipment or other entities
The software process that operated of computing device node and main frame according to the present invention can also be provided.Node according to the present invention
Multiple software process in multiple data communications equipment can also be distributed in main frame or on one group of small, dedicated computer
Between all software process running on all software process running or single computer.
It should be understood that strictly, embodiments of the invention can be implemented as software program in data handling equipment, soft
Part and hardware or single software and/or single circuit.
So far already in connection with preferred embodiment, invention has been described.It should be understood that those skilled in the art are not
In the case of departing from the spirit and scope of the present invention, various other changes, replacement and interpolation can be carried out.Therefore, the present invention
Scope be not limited to above-mentioned specific embodiment, and should be defined by the appended claims.
Claims (8)
1. a kind of data flow heavy route method, including：
According to position in a network, it is switch and host assignment location address, different location address represents different roads
By path；
Select the location address of a routed path and respective host for each data flow, by selected routed path with accordingly
The location address of main frame is stored in stream information table, by the address conversion between the physical address of main frame and the location address of main frame
Rule is stored in the flow table of the edge switch related to main frame；
When newly-increased switch is determined by Topology Discovery, to newly-increased switch and the switch related to newly-increased switch and master
The extra location address of machine distribution, renewal location address table, the routed path in renewal stream information table and the position of respective host
Address translation rule in address, and the flow table of the edge switch updating this data flow；And
Find out all data flows needing heavy-route according to the requirement of load balancing, select new routed path for each data flow
Location address with respective host.
2. data flow heavy route method according to claim 1, one main frame of wherein at least has two location address.
3. data flow heavy route method according to claim 1 and 2, also includes：
Based on the location address being distributed, it is that all switches in network calculate the flow table forwarding for packet, and will count
The flow table calculating is installed in switch.
4. data flow heavy route method according to claim 1, wherein
When generation link failure or exchange fault are determined by Topology Discovery, find out affected all data flows, be
Each data flow selects the location address of new routed path and respective host, updates routed path in stream information table and corresponding
Address translation rule in the location address of main frame, and the flow table of the edge switch updating this data flow.
5. a kind of controller, including：
Switch management unit, for according to position in a network, being switch and host assignment location address, produces position
Address table, different location address represents different routed paths, and described switch management unit is additionally operable to as each data flow
Select the location address of a routed path and respective host, the location address of selected routed path and respective host is deposited
Storage is in stream information table, and the address translation rule between the physical address of main frame and the location address of main frame is notified to give master
The related edge switch of machine, described switch management unit is additionally operable to finding newly-increased switch by checking connection relational table
When, the location address extra to newly-increased switch and the switch related to newly-increased switch and host assignment, more new position ground
Location table, the routed path in renewal stream information table and the location address of respective host, and the edge exchange updating this data flow
Address translation rule in the flow table of machine；
By indicating all switch transmission link layers, Topology Discovery unit, for finding that protocol data bag finds network and opens up
Flutter, produce connection relational table；
Memorizer, for storing connection relational table, location address table and stream information table；And
Heavy-route unit, for, when by checking that connection relational table finds newly-increased switch, finding out wanting according to load balancing
Seek all data flows needing heavy-route, select the location address of new routed path and respective host for each data flow.
6. controller according to claim 5, wherein said switch management unit is that at least one main frame addresses two
Location address.
7. the controller according to claim 5 or 6, wherein
Described switch management unit is additionally operable to based on the location address being distributed, and is that all switches in network calculate and are used for
The flow table that packet forwards, and the flow table calculating is installed in switch.
8. controller according to claim 5, wherein
Described heavy-route unit is used for, when by checking that connection relational table determines generation link failure or exchange fault, finding out
Affected all data flows, select the location address of new routed path and respective host for each data flow, update stream
Routed path in information table and the location address of respective host, and this data flow will be given to the update notification of address translation rule
Edge switch.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210080081.1A CN103326943B (en) | 2012-03-23 | 2012-03-23 | Data flow heavy route method and controller |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210080081.1A CN103326943B (en) | 2012-03-23 | 2012-03-23 | Data flow heavy route method and controller |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103326943A CN103326943A (en) | 2013-09-25 |
| CN103326943B true CN103326943B (en) | 2017-03-01 |
Family
ID=49195483
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210080081.1A Expired - Fee Related CN103326943B (en) | 2012-03-23 | 2012-03-23 | Data flow heavy route method and controller |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103326943B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015109497A1 (en) * | 2014-01-23 | 2015-07-30 | 华为技术有限公司 | Flow table modifying method, flow table modifying device, and openflow network system |
| CN104869178A (en) * | 2014-02-21 | 2015-08-26 | 中兴通讯股份有限公司 | IP address distribution method, controller and gateway device in SDN-EPS |
| WO2016070332A1 (en) * | 2014-11-04 | 2016-05-12 | 华为技术有限公司 | Flow entry processing method and apparatus |
| WO2016074738A1 (en) * | 2014-11-14 | 2016-05-19 | Huawei Technologies Co., Ltd. | Data routing with a machine learning-based routing model |
| CN107534612B (en) * | 2015-07-31 | 2020-07-10 | 深圳市宇轩网络技术有限公司 | Flow table synchronization realization method and forwarding equipment |
| CN106411746A (en) * | 2015-08-03 | 2017-02-15 | 上海宽带技术及应用工程研究中心 | SDN network data transmission system based on Vlan and method thereof |
| CN105450447B (en) * | 2015-11-17 | 2018-05-18 | 上海斐讯数据通信技术有限公司 | A kind of network system and network operation method |
| CN105634817A (en) * | 2016-01-08 | 2016-06-01 | 广州西麦科技股份有限公司 | Network fault automatic detecting system and method based on SDN |
| CN107425987B (en) * | 2017-09-12 | 2019-11-01 | 中国联合网络通信集团有限公司 | A kind of Switch control method, apparatus and system |
| CN110071987B (en) * | 2019-04-29 | 2021-11-19 | 中车青岛四方车辆研究所有限公司 | Gateway address mapping method, gateway and train network system |
| CN110113219A (en) * | 2019-05-29 | 2019-08-09 | 山东浪潮人工智能研究院有限公司 | A kind of Server location information management method |
| CN114465984B (en) * | 2022-04-12 | 2022-08-23 | 浙江中控研究院有限公司 | Address allocation method, system, device and computer readable storage medium based on transmission path |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101667970A (en) * | 2009-10-23 | 2010-03-10 | 杭州华三通信技术有限公司 | Protection switching method and equipment thereof |
| CN102694720A (en) * | 2011-03-24 | 2012-09-26 | 日电（中国）有限公司 | Addressing method, addressing device, infrastructure manager, switchboard and data routing method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7085225B2 (en) * | 2001-09-27 | 2006-08-01 | Alcatel Canada Inc. | System and method for providing detection of faults and switching of fabrics in a redundant-architecture communication system |
-
2012
- 2012-03-23 CN CN201210080081.1A patent/CN103326943B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101667970A (en) * | 2009-10-23 | 2010-03-10 | 杭州华三通信技术有限公司 | Protection switching method and equipment thereof |
| CN102694720A (en) * | 2011-03-24 | 2012-09-26 | 日电（中国）有限公司 | Addressing method, addressing device, infrastructure manager, switchboard and data routing method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103326943A (en) | 2013-09-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103326943B (en) | Data flow heavy route method and controller | |
| US11805045B2 (en) | Selective routing | |
| US7724674B2 (en) | Deadlock free network routing | |
| US10623303B2 (en) | Method of routing data and switch in a network | |
| EP3022877B1 (en) | Extended remote lfa fast reroute | |
| US7231459B2 (en) | Routing scheme based on virtual space representation | |
| KR102123035B1 (en) | Three stage folded clos optimization for 802.1aq | |
| CN105009523B (en) | The method and apparatus quickly re-routed for IP/MPLS | |
| CN104247348A (en) | Chaining of inline services using software defined networking | |
| CN103685035B (en) | Route renewing method, interchanger and system | |
| EP2596603B1 (en) | Ethernet switch and method for routing ethernet data packets | |
| CN107113238A (en) | Equipment, system and method for business chain | |
| US20150256407A1 (en) | Control apparatus, control method thereof, and program | |
| CN100556048C (en) | A kind of vector data communication is set up the method that vector connects on the net | |
| CN102891800A (en) | Scalable forwarding table with overflow address learning | |
| CN107347035A (en) | Method for searching route, device, distribution node, lookup node and Ingress node | |
| CN102891903B (en) | A kind of NAT method and equipment | |
| EP3192221B1 (en) | Method and network node for scalable computer network partitioning | |
| WO2005015834A1 (en) | A method of automatic protection switching | |
| CN103916322A (en) | Method and device for defining look-up system for network element of software-defined network | |
| CN106656808A (en) | Secondary fully-connected interconnection network fault tolerance fully-adaptive routing method | |
| CN100365999C (en) | Design method for grouping service in optical transmission communication network | |
| KR100810727B1 (en) | A Method of Automatic Protection Switching | |
| JP6344005B2 (en) | Control device, communication system, communication method, and program | |
| JP6296736B2 (en) | Route control device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170301 Termination date: 20180323 |