CN106911584A - It is a kind of based on leaf-flow load sharing method of ridge topological structure, apparatus and system - Google Patents
It is a kind of based on leaf-flow load sharing method of ridge topological structure, apparatus and system Download PDFInfo
- Publication number
- CN106911584A CN106911584A CN201510981555.3A CN201510981555A CN106911584A CN 106911584 A CN106911584 A CN 106911584A CN 201510981555 A CN201510981555 A CN 201510981555A CN 106911584 A CN106911584 A CN 106911584A
- Authority
- CN
- China
- Prior art keywords
- congestion
- equipment
- leaf
- value
- path
- 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.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/12—Avoiding congestion; Recovering from congestion
- H04L47/125—Avoiding congestion; Recovering from congestion by balancing the load, e.g. traffic engineering
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/06—Management of faults, events, alarms or notifications
- H04L41/0654—Management of faults, events, alarms or notifications using network fault recovery
- H04L41/0663—Performing the actions predefined by failover planning, e.g. switching to standby network elements
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W28/00—Network traffic management; Network resource management
- H04W28/02—Traffic management, e.g. flow control or congestion control
- H04W28/0289—Congestion control
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W28/00—Network traffic management; Network resource management
- H04W28/02—Traffic management, e.g. flow control or congestion control
- H04W28/08—Load balancing or load distribution
Abstract
The application is related to moving communicating field, more particularly to a kind of based on leaf-flow load sharing method of ridge topological structure, apparatus and system.In the flow load sharing method based on leaf-ridge topological structure, first leaf equipment sends multiple congestion probe messages by mulitpath to the second leaf equipment, and each congestion probe message carries congestion value, the congestion value is used for the Congestion Level SPCC in the path for indicating the congestion probe message to pass through., according to the congestion value per paths, at least one path of selection is to the second leaf equipment sending data for the first leaf equipment.The scheme provided by the application, realizes the load balancing of whole network.
Description
Technical field
The present invention relates to moving communicating field, more particularly to Ethernet flow load sharing.
Background technology
In current Ethernet, in order to reduce influence of the link failure to business, generally by multiple links
Interaction is backed up, when active link breaks down, by flow switch to reserve link, to ensure business
Q＆r.But this mode can not be such that service traffics are forwarded on main/slave link simultaneously, it is impossible to
Effectively using the bandwidth of link.
In order to support flow load sharing, it will usually which multiple physical links of equipment are tied together, composition
One logical links, as shown in Figure 1.Fig. 1 is to constitute one by multiple physical links in the prior art to patrol
Collect the schematic diagram of link.In Fig. 1, server is connected to equipment 1 by two physical links, the two
Physical link composition logical links LAG1.Equipment 1 is connected to the He of equipment 2 further through two physical links
Equipment 3, the two physical links composition logical links LAG2.When by a logical links transmission message
When, according to corresponding load balancing (English：Load balancing) algorithm selects one of them
Physical link sends the message.And when there is physical link failure, then move back the physical link of the failure
Go out logical links so that message will not load balancing on faulty physical link.
Current evolution to using two-layer leaf-ridge (leaf-spine) topological structure sets up data center's net
Network.Fig. 2 is two-layer leaf-spine topological structure schematic diagrames of the prior art.In Fig. 2, leaf (leaf)
Equipment is connected to different ridge (spine) equipment by multiple physical links.The multiple physical link group
Into a logical links.So, the East and West direction flow between leaf equipment and leaf equipment, just by many
Individual physical link, in each spine equipment room load balancing.How effectively each service traffics to be loaded
Share on each physical link in logical links, affect service traffics in above-mentioned data center network
The quality and performance of forwarding.
In order to realize that each service traffics farthest can uniformly share each physics chain in logical links
Road, while ensureing that message is not out of order again, industry proposes static Hashing algorithm.The algorithm is main according to report
The feature field of text, obtains the physical link index under logical links.Therefore, same message characteristic word
Segment value can produce same Hash result.This static Hashing mode, due to same feature field
A plurality of stream can be all hashing onto on same link, therefore occurs that Hash is unbalanced, cause certain physics
Link load is excessive, or even congestion occurs and cause packet loss, and some physical link light loads,
Bandwidth availability ratio is low, as shown in Figure 3.Fig. 3 is to cause Hash using static Hashing algorithm in the prior art
Unbalanced leaf-spine network topologies schematic diagram.
In Fig. 3, leaf equipment by constitute each physical link of logical links be connected to it is different
Spine equipment.Leaf equipment carries out Hash load balancing to the flow of spine equipment on logical links.
If Hash is uneven, there is congestion and even abandons in the flow that can cause to part spine equipment, and reduces
Data center traffic quality and Consumer's Experience.
To solve the unbalanced drawback of link load that static Hashing algorithm Hash inequality is brought, new business
Forwarding chip supports a kind of dynamic load sharing (English：Dynamic load balancing, referred to as：
DLB) algorithm.Namely：In the case where keeping each service message not out of order, according to each thing under logical links
The loading condition of link is managed, selects a physical link of most lightly loaded to send message, such as Fig. 4
It is shown.
Fig. 4 is that the leaf-spine topologys for carrying out local load balancing using DLB algorithms in the prior art are shown
It is intended to.In Fig. 4, when server 1 asks to send data to server 3, leaf equipment 1 arrives spine
The link 1 of equipment 1 overload, and leaf equipment 1 is loaded to the link 2 of spine equipment 2
It is lighter.Now leaf equipment 1, can be new by this when using DLB algorithms to new business flow load sharing
Service traffics are shared on the light link 2 of load.And if according to static Hashing algorithm, new business flow
Still may proceed to share on the excessive link 1 of load.The final effect of DLB algorithms, can cause logic
Local each physical link load of link is more balanced.
DLB algorithms can very well reach local each physical link in the local load balancing of the equipment of execution
Load balancing effect, same physics chain is all hashing onto without static Hashing algorithm each stream like that
Lu Shang, and cause the physical link congestion and packet loss.Even if but DLB algorithms accomplish that equipment is local
Load balancing, cannot also meet the traffic sharing requirement in data center network leaf-spine topologys.
In leaf-spine topological structures, such as the East and West direction stream of leaf equipment to leaf equipment
Amount, can carry out flow load sharing by multiple physical links and through multiple spine equipment.Such as Fig. 5 institutes
Show, Fig. 5 is to cause the unbalanced leaf-spine topologys of offered load using DLB algorithms in the prior art
Schematic diagram.
In Fig. 5, server 1 ask to server 3 send data when, flow from leaf equipment 1 to
There are two paths up to leaf equipment 3, be respectively made up of physical link { 1,3 }, { 2,4 }.Reach leaf
The light load of link 3 of equipment 3, but has there is congestion discarding state in link 4.Leaf equipment 1 is being held
During row DLB algorithms, the load balancing between local physical link 1 and 2 can only be ensured.If flow is breathed out
Wish on physical link 2, leaf equipment 3 will be reached by link 4, so as to message occur due to gathering around
Fill in and be dropped.It can be seen that, multiple paths are had in network topology can reach the purpose equipment of message,
The link bandwidth of equipment and load weight differ on each paths.Even if DLB algorithms have reached local device
The load balancing of each physical link, cannot also ensure whole path under network topology can load balancing,
Abandoned without congestion.
Therefore, no matter static Hashing algorithm or DLB algorithms are being used, can not all realizing leaf-spine
The flow load sharing in whole path under network topology.
The content of the invention
There is provided herein a kind of based on leaf-flow load sharing method of ridge topological structure, apparatus and system,
To realize the traffic load balancing of whole network.
On the one hand, embodiments herein provides a kind of flow load sharing side based on leaf-ridge topological structure
Method.Wherein, the first leaf equipment is in communication with each other by ridge equipment to the second leaf equipment.The method includes should
First leaf equipment sends multiple congestion probe messages, each congestion by mulitpath to the second leaf equipment
Probe message includes congestion value field.The congestion value field is currently passed through for writing the congestion detection information
Equipment go the congestion value on top, the congestion value to be used to indicate the Congestion Level SPCC in path.The first leaf equipment connects
Multiple response messages of the second leaf equipment transmission are received, wherein each response message includes the response message pair
The congestion value in the path that the congestion probe message answered is passed through.The first leaf equipment is according to the plurality of response message
It is determined that the congestion value per paths, selects at least one path as to second leaf from the mulitpath
The path of equipment sending data.
The present invention realizes the flow load sharing of leaf-spine whole networks, it is ensured that data center's net
The transmission performance and handling capacity of Business Stream, increased network bandwidth utilization factor in network, improve Consumer's Experience.
In a possible design, the congestion value per paths is the physics chain included according to the path
What the congestion value on road determined.Further, the congestion value of the physical link is according to physical link correspondence
What the quantized value of the quantized value of the queue length of exit port and/or the bandwidth availability ratio of correspondence exit port determined.
In a possible design, congestion detection disappears the quantized value of the quantized value bandwidth availability ratio of queue length
Cease is protocol massages or data message, the congestion probe message also mark including path.The congestion value and
The mark in the path is included in the frame head of the congestion probe message, reserved bit or new field.
In a possible design, the flow load sharing method also includes that the first leaf equipment receives second
The reverse blockage probe message that leaf equipment sends.Obtained from the congestion value field of the reverse blockage probe message
Take the reverse blockage value in the path of reverse blockage control message process.
On the other hand, the embodiment of the invention provides a kind of based on leaf-ridge topological structure flow load sharing
Leaf equipment.The leaf equipment is in communication with each other with another leaf equipment.The leaf equipment include transmitter, receiver and
Processor.
The transmitter sends multiple congestion probe messages by mulitpath to another leaf equipment, and each is gathered around
Plug probe message includes congestion value field, and the congestion value field is currently passed through for writing the congestion probe message
The congestion value that the equipment crossed determines, the congestion value is used to indicate the Congestion Level SPCC in path.The receiver is used for
Multiple response messages that another leaf equipment sends are received, wherein each response message includes the response message
The congestion value in the path that corresponding congestion probe message is passed through.The processor is used to disappear according to the plurality of response
Breath determine per paths congestion value, selected from the mulitpath at least one path as to this second
The path of leaf equipment sending data.
Another aspect, the embodiment of the invention provides a kind of based on leaf-ridge topological structure flow load sharing
Ridge equipment.Wherein, the first leaf equipment is in communication with each other by ridge equipment by ridge equipment with the second leaf equipment.
The ridge equipment includes receiver and processor.The receiver is used to receive the congestion of the first leaf equipment transmission
Probe message, the congestion value field of the congestion probe message includes the first thing of the first leaf equipment to ridge equipment
Manage the congestion value of link.The processor is used for congestion value and ridge equipment by comparing first physical link
To the second leaf equipment congestion value determine the first leaf equipment via ridge equipment gathering around to the path of the second leaf equipment
Plug value, the congestion value according to the determination processes the congestion probe message.
In a possible design, when congestion probe message is processed according to the congestion value of the determination, should
Processor specifically for：When the congestion value of the first physical link is the congestion value in path, directly to second
Leaf equipment sends congestion probe message；When the congestion value of the second physical link is the congestion value in path, will
The congestion value of the first physical link in the congestion value field of congestion probe message is updated to the second physical link
Congestion value, and send the congestion probe message after the renewal to the second leaf equipment.
Another aspect, the embodiment of the invention provides a kind of based on leaf-ridge topological structure flow load sharing
System.First leaf equipment is in communication with each other via ridge equipment with the second leaf equipment.The system sets including the first leaf
Standby, ridge equipment, the second leaf equipment.
The first leaf equipment is used to send congestion probe message, and the congestion probe message bag to the ridge equipment
Include congestion value field.The congestion value field is used to write the congestion value of the first leaf equipment determination, the congestion
It is worth for indicating the first leaf equipment to the first physical link Congestion Level SPCC of the ridge equipment.
The ridge equipment will be used to compare the first physical link congestion value with the ridge equipment to the second leaf equipment
The second physical link congestion value determine the first leaf equipment via the ridge equipment to the road of the second leaf equipment
The congestion value in footpath.
The second leaf equipment is used to receive the congestion value in the path from the ridge equipment, and the path is gathered around
Plug value feeds back to the first leaf equipment, and choosing respective paths so as to the first leaf equipment sends data.
Compared to prior art, present invention optimizes the load point of multipath under leaf-spine network topologies
Load effect, it is to avoid service traffics are shared onto the forward-path of obstruction, improve transfer quality and performance,
Improve network bandwidth utilization factor.And in physical link failure, the present invention can automatic sensing and from
Dynamic adjustment is shared to Business Stream, improves robustness of the data center network to physical link failure.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be right
The accompanying drawing to be used needed for embodiment or description of the prior art is briefly described.
Fig. 1 is to constitute a schematic diagram for logical links by multiple physical links in the prior art；
Fig. 2 is two-layer leaf-spine topological structure schematic diagrames of the prior art；
Fig. 3 is to cause the unbalanced leaf-spine networks of Hash using static Hashing algorithm in the prior art
Topological schematic diagram；
Fig. 4 is that the leaf-spine topologys for carrying out local load balancing using DLB algorithms in the prior art are shown
It is intended to；
Fig. 5 is to cause the unbalanced leaf-spine topologys of offered load using DLB algorithms in the prior art
Schematic diagram；
Fig. 6 is the leaf-spine involved by flow load sharing method provided in an embodiment of the present invention
System architecture schematic diagram；
Fig. 7 is the flow load sharing side based on leaf-spine topological structures provided in an embodiment of the present invention
Method flow chart；
Fig. 8 is the leaf based on leaf-spine topological structure flow load sharings provided in an embodiment of the present invention
Equipment schematic diagram；
Fig. 9 is the ridge based on leaf-spine topological structure flow load sharings provided in an embodiment of the present invention
Equipment schematic diagram；
Figure 10 is provided in an embodiment of the present invention based on leaf-spine topological structure flow load sharing systems
Schematic diagram.
Specific embodiment
Below in conjunction with accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely retouched
State.
Fig. 6 is that the flow based on leaf-spine network topology structures provided in an embodiment of the present invention is born
Carry the leaf-spine system architecture schematic diagrams involved by sharing method.The system includes multiple leaf
Equipment, multiple spine equipment, multiple servers.Connected entirely between leaf equipment and spine equipment,
Ensure the reliability of network.It is not connected between spine equipment and spine equipment, leaf equipment and leaf
It is not connected between equipment.
Spine equipment 1, spine equipment 2 are the backbone nodes of bearer network, positioned at convergence-level.leaf
1~leaf of equipment equipment 4 is the leaf node of bearer network, positioned at Access Layer.
In one example, spine equipment is a kind of Ethernet switch, leaf equipment be also it is a kind of with
Too network switch.Further, interactive machine performance of the switch performance of spine equipment than leaf equipment
It is stronger.For example, the interchanger of spine equipment supports 100Gb/s ports, the interchanger branch of leaf equipment
Hold 1Gb/s or 10Gb/s ports.
It will be understood by those skilled in the art that the leaf-spine network structures in Fig. 6 do not constitute it is right
The restriction of leaf-spine network structures, leaf equipment, spine number of devices can than illustrate it is more or
Less.
In Fig. 6, when server 1 sends data to server 3, there are two paths available, first
Paths are path { 1,3 }, i.e., first through physical link 1 again through physical link 3.Second paths are { 2,4 },
I.e. first through physical link 2 again through physical link 4.Physical link 1 is set from leaf equipment 1 to spine
Standby 1 link, physical link 3 is the link from spine equipment 1 to leaf equipment 3.Physical link
2 is the link from leaf equipment 1 to spine equipment 2, physical link 4 be from spine equipment 2 to
The link of leaf equipment 3.In order to choose a light path of Congestion Level SPCC, the embodiment of the invention provides
It is a kind of local to realize leaf equipment according to the path load Congestion Level SPCC of leaf-spine whole networks
Load balancing, it is achieved thereby that global load balancing (English：Global load balancing, GLB).
It is described below how the embodiment of the present invention realizes GLB.
In Fig. 6, before server 1 sends data to server 3, the leaf that server 1 is accessed
Equipment 1 for detective path { 1,3 }, the Congestion Level SPCC of path { 2,4 }, to path { 1,3 }, path { 2,4 }
Congestion probe message is sent respectively, and the congestion probe message includes congestion value field, and the congestion value field is used
In the Congestion Level SPCC quantized value for writing the equipment determination that the congestion probe message is currently passed through.Referred to as should herein
Congestion Level SPCC quantized value is congestion value, and the congestion value is used to indicate the Congestion Level SPCC in path.The path is gathered around
The congestion value of each i.e. each physical link of physical link Congestion Level SPCC quantized value that plug value is included according to the path
It is determined that, and every congestion value of physical link is then calculated by the end equipment of the physical link, i.e., by leaf
Equipment and/or spine equipment are calculated the congestion value of its corresponding physical link.The basis of leaf equipment 1
The path { 1,3 } of determination, the congestion value of path { 2,4 }, choose the light path of Congestion Level SPCC, that is, choose and gather around
The small path of plug value sends the data from server 1, so as to realize the complete of whole leaf-spine networks
Office's load balancing.
Fig. 7 is the global load point based on leaf-spine network topology structures provided in an embodiment of the present invention
Load method flow diagram.
Each equipment in step 710, leaf-spine networks is determined with the equipment as end equipment
Every congestion value of physical link.
Each equipment (including leaf equipment and spine equipment) in leaf-spine network topology structures
Congestion Level SPCC to every physical link with the equipment as end equipment quantifies, each so as to obtain
The congestion value of leaf equipment, every physical link of spine equipment.
For example, in Fig. 6, leaf equipment 1 is distinguished the Congestion Level SPCC of physical link 1, physical link 2
Quantified.Leaf equipment 3 quantifies respectively to the Congestion Level SPCC of physical link 3, physical link 4.
Spine equipment 1 quantifies respectively to the Congestion Level SPCC of physical link 1 ', physical link 3 '.spine
Equipment 2 quantifies respectively to the Congestion Level SPCC of physical link 2 ', physical link 4 '.Wherein, physics chain
Road 1 ' is link of the spine equipment 1 to leaf equipment 1, physical link 3 ' be spine equipment 1 to
The link of leaf equipment 3；Physical link 2 ' is link of the spine equipment 2 to leaf equipment 1, thing
Reason link 4 ' is link of the spine equipment 2 to leaf equipment 3.
Illustrate leaf equipment, spine equipment how to the leaf equipment or spine equipment in detail below
For the Congestion Level SPCC of the physical link of end equipment is quantified, to obtain every congestion value of physical link.
In one example, leaf equipment, spine equipment are carried out to every Congestion Level SPCC of physical link
Quantify, be according to the team to physical link correspondence exit port so as to obtain the congestion value of respective physical link
Row length is quantified and/or the bandwidth availability ratio that local physical link corresponds to exit port is carried out to quantify really
Fixed.
The exit port queue length refers to that the exit port that leaf equipment, spine device periodically read out is delayed
Queue length is deposited, wherein, the length of buffer queue refers to message byte number in buffer queue.
The physical link bandwidth availability ratio refers to the physics that leaf equipment, spine device periodically read out
Message byte number, i.e. service traffics transmitted by the corresponding exit port of link, exit port corresponding to the physical link
Bandwidth, such as with the port of a width of 10G, percentage.
In one example, leaf equipment or spine equipment are to physical link queue length and/or physics chain
The scope of road bandwidth availability ratio carries out segment quantization, one quantized value of correspondence per segment limit, and by chart
Mode stores in relevant device buffering area quantized value corresponding to each scope, referring to table 1 below, 2.For example,
Leaf equipment 1 is stored in its buffering area and represents each segment limit quantized value of its physical link queue length
Form (referring to table 1), and store expression each segment limit quantized value of its physical link bandwidth availability ratio
Form (referring to table 2).
The exit port congestion value can be according to the quantized value of the exit port queue length and/or the exit port bandwidth
Utilization rate determine.Illustrate how to quantify exit port queue length below by table 1 below, pass through
Table 2 below illustrates how to quantify exit port bandwidth availability ratio.
Table 1 below is quantized into as a example by 3bit (bit) by by the exit port queue length, to the exit port
Queue length is quantified.
Table 1
In upper table 1, such as certain leaf equipment or spine equipment exit ports queue length are 20000 words
Section, then the leaf equipment or spine equipment is by inquiring about upper table 1, the leaf equipment or spine
The exit port queue length of equipment is quantized into 6, that is, be quantized into the 110 of 3bit.
It should be noted that exit port queue length in above-mentioned table 1 quantify waypoint, quantizing range,
Concrete numerical value in quantized result, user be can configure, namely concrete numerical value is not limited to shown in table 1.
Table 2 below is quantized into as a example by 3bit by by the exit port bandwidth availability ratio, to exit port bandwidth profit
Quantified with rate.
Table 2
In upper table 2, such as certain leaf equipment or spine equipment exit port bandwidth availability ratios are 86%,
Then, by inquiring about upper table 2, the leaf equipment or spine set for the leaf equipment or spine equipment
Standby exit port bandwidth availability ratio is quantized into 5, that is, be quantized into the 101 of 3bit.
It should be noted that exit port bandwidth availability ratio in above-mentioned table 2 quantify waypoint, quantizing range,
Concrete numerical value in quantized result, user be can configure, namely concrete numerical value is not limited to shown in table 2.
Additionally, to improve quantified precision, it is also possible to which the bit wide of above-mentioned table 1 and/or the quantized result of table 2 is increased
Plus, waypoint is quantified in such cases also accordingly to be increased.It is more to quantify bit wide, it is meant that to produce more preferable
Quantization accuracy, and preferably reflect physical link Congestion Level SPCC.
In one example, in leaf-spine networks, leaf equipment or spine equipment are by searching
The exit port queue length quantifies table (i.e. table 1) and/or the exit port bandwidth availability ratio quantifies table (i.e. table
2) mode, obtains every congestion value of physical link.Congestion value is bigger, represents gathering around for the physical link
Plug degree is higher.
For example, leaf equipment or spine equipment can by the quantized value to exit port queue length and
Quantized value to exit port bandwidth availability ratio assigns the mode of weight respectively, obtains the leaf equipment or spine
The congestion value of equipment physical link.And for example, leaf equipment or spine equipment can be by by the exit ports
The mode that the quantized value of queue length is added or is multiplied with the quantized value of the exit port bandwidth availability ratio, obtains
To the leaf equipment or spine equipment physical link congestion values.For another example, leaf equipment or spine equipment
The leaf equipment or spine equipment physics chains only can be worth to by the quantization of the exit port queue length
Road congestion value；Or can only by the quantization of the exit port bandwidth availability ratio be worth to the leaf equipment or
Spine equipment physical link congestion values.
Continue with other steps in elaboration Fig. 7.
Step 720 is to what step 730 was illustrated：Server 1 is pre- to send data to server 3.First
Leaf equipment 1 (i.e. the first leaf equipment) corresponding to server 1 is by path { 1,3 } to leaf equipment
3 send congestion probe message 1, while leaf equipment 1 is sent by path { 2,4 } to leaf equipment 3
Congestion probe message 2.The congestion probe message 1 includes congestion value field 1, and the congestion value field 1 is used for
The congestion value that write-in spine equipment 1 determines, the congestion value is used to indicate the Congestion Level SPCC of path { 1,3 }.
The congestion probe message 2 includes congestion value field 2, and the congestion value field 2 is used to write spine equipment 2
The congestion value of determination, the congestion value is used to indicate the Congestion Level SPCC of path { 2,4 }.
In step 720, the first leaf equipment sends multiple congestion probe messages, each congestion by physical link
Probe message includes congestion value field.For example, leaf equipment 1 passes through physical link 1 to spine equipment
1 sends congestion probe message 1, and the congestion probe message 1 includes congestion value field 1, the congestion value field
1 includes the congestion value of the physical link 1 that leaf equipment 1 determines, the congestion value of the physical link 1 is used for
Indicate the Congestion Level SPCC of the physical link 1；Leaf equipment 1 is set by physical link 2 to spine simultaneously
Standby 2 send congestion probe message 2, and the congestion probe message 2 includes congestion value field 2, the congestion value word
Section 2 includes the congestion value of the physical link 2 that the leaf equipment 1 determines, the congestion value of the physical link 2
Congestion Level SPCC for indicating the physical link 2.
In one example, leaf equipment 1 is by searching its exit port queue length quantization table (i.e. table 1)
And/or the mode of its exit port bandwidth availability ratio quantization table (i.e. table 2) obtains physical link 1 and physics chain
The congestion value on road 2.Concrete methods of realizing is described referring to above-mentioned steps 710 and corresponding contents.Show at one
In example, the congestion probe message is protocol massages or data message, and the congestion probe message also includes path
Mark, the mark in the congestion value and the path be included in the frame head of the congestion probe message, reserved bit or
In new field.Below so that the frame head of congestion probe message includes congestion value, ID of trace route path as an example, carry out
Illustrate.
The Congestion Level SPCC in each path determines according to the Congestion Level SPCC metric of its physical link for being included.
The metric of the physical link Congestion Level SPCC includes that the physical link congestion value, physical link are identified.Ginseng
See the table below 3.
| 4bit | 6bit |
| Path-Id | Path-Congestion-Quantized-Value |
Table 3
In table 3, the physical link Congestion Level SPCC metric that the congestion probe message frame head includes is included：
Physical link mark Path-Id, the congestion value of physical link
Path-Congestion-Quantized-Value.Wherein, the Path-Id can be represented by 4bit,
The congestion value Path-Congestion-Quantized-Value of physical link can be by 6bit (bit)
Represent.
It will be understood by those skilled in the art that Path-Id bit wides are not defined to concrete numerical value 4bit,
And Path-Congestion-Quantized-Value bit wides are not defined to concrete numerical value 6bit yet.
When Path-Id bit wides are 4bit, represent that leaf equipment can at most there are 16 physical links to reach spine
Equipment.If in leaf-spine topological structures, there is more spine equipment, i.e. spine number of devices
When amount is more than 16, then Path-Id bit wides are needed to be more than 4bit.Path-Id bit wide actual values are according to leaf
The quantity that equipment reaches spine equipment physical links is determined.
In step 730, ridge equipment determines each physical link that the path of the congestion probe message for receiving includes
Extreme congestion value, and according to the extreme congestion value treatment congestion probe message.For example, spine sets
Standby 1 receives the congestion probe message 1 from leaf equipment 1, identifies that the congestion probe message 1 is hair
Toward the congestion probe message of leaf equipment 3, then the spine equipment 1 determines the congestion probe message
Path also includes physical link 3.Then the spine equipment 1 compares the frame head of congestion probe message 1 and includes
Congestion value and physical link 3 congestion value.If the congestion value of the physical link 3 is visited more than the congestion
The congestion value that message 1 includes is surveyed, then updates the congestion probe message 1 with the congestion value of the physical link 3
The congestion value that frame head includes, does not update otherwise, and directly being sent to leaf equipment 3 (the second leaf equipment) should
Congestion probe message 1.That is, spine equipment 1 chooses physical link 1, physical link 3 gathering around
The maximum of plug value and is carried as the congestion value of path { 1,3 } by the frame head of the congestion probe message 1
The congestion value in the path { 1,3 }.Then this is included gathering around for the congestion value of path { 1,3 } by spine equipment 1
Plug probe message 1 is sent to leaf equipment 3.
In the identical mode of spine equipment 1, spine equipment 2 will be including the congestion value of path { 2,4 }
Congestion probe message 2 is also sent to leaf equipment 3.
It should be noted that above-mentioned steps 730 are the extreme congestions by each physical link in path selection
It is worth as the congestion value in the path, to weigh the Congestion Level SPCC in the path.In fact, not limited to this,
Can be by by the mode such as each physical link quantized value addition or multiplication in path, obtaining the congestion in the path
Value, to weigh the Congestion Level SPCC in the path.
In step 740, the second leaf equipment obtains congestion value and stores from each congestion probe message.For example,
Leaf equipment 3 receives the congestion probe message 1 from leaf equipment 1, congestion probe message 2.leaf
Equipment 3 extracted respectively from the congestion probe message 1, the frame head of congestion probe message 2 Path-Id and
The mark of the value of Path-Congestion-Quantized-Value fields, i.e. acquisition approach and the road
The congestion value in footpath, and the congestion value in the mark in the path and the path stored to the leaf equipment 3
In " From Leaf " congestion table in.Being somebody's turn to do " From Leaf " congestion table is used to store leaf equipment 3
The mark in the path of the congestion probe message that other leaf equipment received as purpose equipment send, and institute
State the congestion value in path.
It should be noted that identifying Path-Id by physical link, it becomes possible to obtain corresponding ID of trace route path.
For example, leaf equipment 3 obtains physical link mark by the field Path-Id in congestion probe message frame head
It is 1 to know, and learns only had by the spine equipment of leaf equipment 1 to leaf equipment 3 by physical link 1
Spine equipment 1, according to spine equipment 1 and source device leaf equipment 1, purpose equipment leaf equipment
3 would know that respective paths are designated { 1,3 }.
In step 750, the second leaf equipment disappears the congestion value in the path of each congestion probe message by response
Breath feeds back to the first leaf equipment.Wherein described response message can be reverse blockage probe message.For example,
After leaf equipment 3 receives congestion probe message 1 and congestion probe message 2, you can know path { 1,3 },
The congestion value in path { 2,4 }.However, now, only purpose equipment leaf equipment 3 knows every paths
Congestion value, and source device leaf equipment 1 is not aware that.Therefore, leaf equipment 3 is by path { 1,3 }
Congestion value by reverse blockage probe message 3, and feed back to leaf equipment 1 via path { 3 ', 1 ' }；
Simultaneously also by the congestion value of path { 2,4 } by reverse blockage probe messages 4, and via path { 4 ', 2 ' }
Feed back to leaf equipment 1.Wherein, path { 3 ', 1 ' } refers to first to arrive physical link again through physical link 3 '
1 ' path, and the expression of physical link 3 ' is by the physical link of leaf equipment 3 to spine equipment 1, thing
Reason link 1 ' is represented by the physical link of spine equipment 1 to leaf equipment 1；Path { 4 ', 2 ' } refers to
First arrive the path of physical link 2 ' again through physical link 4 ', and physical link 4 ' is represented by leaf equipment 3
To the physical link of spine equipment 2, the expression of physical link 2 ' is by spine equipment 2 to leaf equipment 1
Physical link.
In one example, in order to economize on resources, the quantity of congestion probe messages is reduced, in leaf equipment
3 by during the congestion value of path { 3 ', 1 ' } to the feedback path of leaf equipment 1 { 1,3 }, measuring
The congestion value in path { 3 ', 1 ' }；Equally, set to leaf by path { 4 ', 2 ' } in leaf equipment 3
During the congestion value of standby 1 feedback path { 2,4 }, the congestion value of measuring route { 4 ', 2 ' }.In other words
Say, leaf equipment 3 is in path { 3 ', 1 ' }, { 4 ', 2 of the equipment 1 of measurement leaf equipment 3 to leaf
' congestion value when, road is carried by the frame head (or reserved bit or new field) of congestion probe message respectively
Footpath { 1,3 }, the congestion value of { 2,4 }.
Therefore, the frame head of the congestion probe message is except including for storing the congestion determined in measurement process
Quantized result：Path-Id (physical link mark), Path-Congestion-Quantized-Value
Outside (congestion value) field, also add the Path-Id-Metric physical link of return (mark),
Path-Congestion-Quantized-Value-Metric (the congestion value of return) field, for taking
Band returns to the congestion value in the path of source leaf equipment.Referring specifically to table 4 below
Table 4
It will be understood by those skilled in the art that Path-Id, Path-Id-Metric bit wide in table 4 are simultaneously
It is not defined to concrete numerical value 4bit, and Path-Congestion-Quantized-Value,
Path-Congestion-Quantized-Value-Metric bit wides are not also defined to concrete numerical value
6bit。
In step 760, the first leaf equipment receives multiple response messages, obtains bag in each response message
The congestion value in the path for including, selects at least one path as to second leaf from the mulitpath
The path of equipment sending data.The response message can be reverse blockage probe message.For example, leaf
Equipment 1 receives multiple response messages that leaf equipment 3 sends, and wherein each response message includes the response
The congestion value in the path that the corresponding congestion probe message of message is passed through.Leaf equipment 1 is according to the plurality of response
Message determine per paths congestion value, and selected from the mulitpath at least one path as to
Leaf equipment 3 sends the path of data.
Specifically, leaf equipment 1 receives reverse blockage probe message 3 from leaf equipment 3 and reversely
Congestion probe message 4, the congestion value of acquisition approach { 1,3 } from the frame head of reverse blockage probe message 3,
The congestion value of acquisition approach { 2,4 } from the frame head of reverse blockage probe message 4.Leaf equipment 1 compares road
Footpath { 1,3 } congestion value and path { 2,4 } congestion value, choose congestion and are worth small path, will be from server 1
Data sent to server 3 by the path.
In one example, when the reverse blockage detection that leaf equipment 1 receives the transmission of leaf equipment 3 disappears
During breath, that is, when receiving reverse blockage probe message 3, reverse blockage probe message 4, except from the message
Reverse blockage degree metric, including the Path-Id-Metric (things of return are extracted and preserved in frame head
Reason link identification), Path-Congestion-Quantized-Value-Metric (the congestion value of return)
Outside, also extracted from the frame head of the reverse blockage probe message and preserved by leaf equipment 3 to leaf
The Congestion Level SPCC metric of equipment 1, including Path-Id (physical link mark),
Path-Congestion-Quantized-Value (congestion value).
The reverse blockage degree metric that leaf equipment 1 is extracted includes two groups, and one group is path { 1,3 }
Mark and its congestion value, another group be path { 2,4 } mark and its congestion value.Leaf equipment 1 is extracted
Stored to " To Leaf " congestion table after going out two groups of reverse blockage degree metrics, should " To
The storage of Leaf " congestions table is that leaf equipment 1 sends congestion spy as source device to other leaf equipment
Survey message path, and the path congestion value.Leaf equipment 1 extract by leaf equipment 3 to
The Congestion Level SPCC metric of leaf equipment 1 also include two groups, one group be path { 3 ', 1 ' } mark and
Its congestion value, another group be path { 4 ', 2 ' } mark and its congestion value.Leaf equipment 1 extracts this
Stored after two groups of Congestion Level SPCC metrics to " From Leaf " and quantified in congestion table, should " From
The storage of Leaf " congestions table be leaf equipment 1 as purpose equipment receive from other leaf equipment
The mark in the path of the congestion probe message of transmission, and the path congestion value.
In the same fashion, all leaf equipment, the spine equipment pair in leaf-spine network structures
Congestion Level SPCC per paths is measured, and final each leaf equipment, spine equipment form respective gathering around
Plug table.When load balancing is performed to Business Stream, " To Leaf " congestion that leaf equipment is stored according to it
For the congestion value in each path corresponding to certain specific purpose leaf equipment in table, congestion value minimum is chosen
Path sends data.
It should be noted that described above is a reverse blockage probe message frame head, except including its institute
Outside the congestion value in the path of process, only including a reverse blockage value.For example, reverse blockage detection disappears
Cease 3 frame heads to remove outside the congestion value for including path { 3 ', 1 ' }, only including the congestion value of path { 1,3 }.
In fact, a reverse blockage probe message frame head can include multiple reverse blockage values.For example, reversely
Congestion probe message 3 can include multiple reverse blockage values, such as the congestion value of path { 1,3 } and path { 2,4 }
Congestion value.
Additionally, described above is by congestion probe message (or reverse blockage probe message) frame head
Include congestion value.In fact, congestion value can also be carried by other means.For example, passing through business
Message carries congestion value.In such cases, it is necessary to leaf equipment and spine equipment support that service message can
Extra to increase field, leaf equipment carries the congestion value by the extra increased field, leaf equipment
After congestion value is got, the extra increased field is divested from the service message.And for example, pass through
The reserved bit (Res) of service message field transmits the congestion value.
To sum up, by the GLB load balancings of the embodiment of the present invention, leaf equipment 1 sends purpose to Leaf
The service message of equipment 3 can get around distal end physical link and the path of congestion occurs, it is ensured that Business Stream
Transmission performance and handling capacity, improve Consumer's Experience, increased the utilization rate of the network bandwidth.
Fig. 8 is the leaf based on leaf-spine topological structure flow load sharings provided in an embodiment of the present invention
Equipment schematic diagram.The leaf equipment includes transmitter 810, receiver 820 and processor 820.
Transmitter 810 is used to send multiple congestion probe messages to another leaf equipment by mulitpath, often
Individual congestion probe message includes congestion value field, and the congestion value field is used to indicate the Congestion Level SPCC in path.
In one example, the path congestion value is the congestion value of the physical link included according to the path
Determine.Further, the path congestion value be the path the maximum comprising each physical link congestion value
Value.For example, in Fig. 6, the congestion value of path { 1,3 } is physical link 1, the congestion value of physical link 3
Maximum.
In one example, the physical link congestion value is the queue according to physical link correspondence exit port
What the quantized value of the bandwidth availability ratio of the quantized value of length and/or physical link correspondence exit port determined.Tool
Body method refers to above-mentioned steps 710 and its corresponding contents description.
Receiver 820 is used to receive multiple response messages that another leaf equipment sends, and wherein each response disappears
Breath includes the congestion value in the path that the corresponding congestion probe message of the response message is passed through.
In one example, the response message that receiver 820 is received is above-mentioned reverse blockage probe message,
I.e. from the congestion probe message of purpose leaf equipment, the message includes the leaf equipment to another leaf equipment
The congestion value in path.For example, in Fig. 6, reverse blockage probe message 3 includes the congestion value of path { 1,3 }.
Further, the reverse blockage probe message that the leaf equipment is received also includes that another leaf equipment extremely should
The congestion value of the leaf equipment.For example, in Fig. 6, reverse blockage probe message 3 also includes path { 3 ', 1 ' }
Congestion value.Processor 830 is used to determine every road according to the multiple response messages from receiver 820
The congestion value in footpath, selects at least one path to send number as to another leaf equipment from the mulitpath
According to path.
In one example, the congestion probe message is protocol massages or data message, and congestion detection disappears
The breath also mark including path.The mark in the congestion value and the path is included in the frame of the congestion probe message
In head or reserved bit or new field.
In one example, the leaf equipment chooses the path of congestion value minimum in the plurality of path, to above-mentioned
Another leaf equipment sending data.
Fig. 9 is the ridge based on leaf-spine topological structure flow load sharings provided in an embodiment of the present invention
Equipment schematic diagram.In Fig. 9, the first leaf equipment is in communication with each other by the ridge equipment with the second leaf equipment.
The ridge equipment includes receiver 910 and processor 920.
Receiver 910 is used to receive the congestion probe message of the first leaf equipment transmission, and the congestion is detected
The congestion value field of message includes the first physical link congestion value of the first leaf equipment to the ridge equipment.
In one example, the congestion probe message is protocol massages or data message, and congestion detection disappears
The breath also mark including path.The mark in the congestion value and the path is included in the frame of the congestion probe message
In head or reserved bit or new field.
In one example, the physical link congestion value is long according to the queue of physical link correspondence exit port
What the quantized value of the bandwidth availability ratio of the quantized value of degree and/or physical link correspondence exit port determined.Specifically
Method refers to above-mentioned steps 710 and its corresponding contents description.
Processor 920 is used to be set with the ridge equipment to the second leaf by comparing the first physical link congestion value
The second standby physical link congestion value, determines the first leaf equipment via the ridge equipment to the second leaf equipment
Path congestion value, the congestion value according to the determination processes the congestion probe message.
In one example, processor 920 is by the first physical link congestion value and second physical link
The maximum of congestion value as the first leaf equipment via the ridge equipment to the second leaf equipment path congestion
Value.
In one example, when the congestion probe message is processed according to the congestion value of the determination, processor
920 specifically for：When the congestion value that the congestion value of the first physical link is the path, directly to this second
Leaf equipment sends the congestion probe message；When the congestion value that the congestion value of second physical link is the path
When, the congestion value of first physical link in the congestion value field of the congestion probe message is updated to this
The congestion value of the second physical link, and send the congestion probe message after the renewal to the second leaf equipment.
Figure 10 is provided in an embodiment of the present invention a kind of to be based on leaf-ridge topological structure flow load sharing
System.The system includes the first leaf equipment 101, ridge equipment 102, the second leaf equipment 103.First leaf equipment
101 are in communication with each other via ridge equipment 102 with the second leaf equipment 103.
First leaf equipment 101 is used to send congestion probe message to ridge equipment 102, and congestion detection disappears
Breath includes congestion value field.The congestion value field is used to write the congestion value of the first leaf equipment determination, should
Congestion value is used to indicate the first leaf equipment 101 to the first physical link Congestion Level SPCC of ridge equipment 102.
Ridge equipment 102 is used to compare the first physical link congestion value and the leaf equipment of ridge equipment 102 to the second
103 the second physical link congestion value, determines that the first leaf equipment 101 sets via the leaf of ridge equipment 102 to the second
The congestion value in standby 103 path.
Second leaf equipment 103 is used to receive the congestion value in the path from ridge equipment 102, Bing Jianggai roads
Footpath congestion value feeds back to the first leaf equipment 101, and choosing respective paths so as to the first leaf equipment 101 sends data.
In one example, the physical link congestion value is the queue according to physical link correspondence exit port
What the quantized value of the quantized value of length and/or the bandwidth availability ratio of correspondence exit port determined.
In one example, the congestion probe message is protocol massages or data message, and congestion detection disappears
The breath also mark including path.The mark in the congestion value and the path is included in the frame of the congestion probe message
In head or reserved bit or new field..
In one example, the path congestion value is the first physical link congestion value and the second physical link
The maximum of congestion value.
Above-described specific embodiment, is carried out to the purpose of the present invention, technical scheme and beneficial effect
Further describe, should be understood that the foregoing is only specific embodiment of the invention and
, the protection domain being not intended to limit the present invention, it is all on the basis of technical scheme,
Any modification, equivalent substitution and improvements done etc., all should be included within protection scope of the present invention.
Claims (15)
1. a kind of flow load sharing method based on leaf-ridge topological structure, wherein, the first leaf equipment passes through
Ridge equipment is in communication with each other with the second leaf equipment, and methods described includes：
The first leaf equipment is sent multiple congestions and detected and disappeared by mulitpath to the second leaf equipment
Breath, each congestion probe message includes congestion value field, and the congestion value field is used to write the congestion
The congestion value that the equipment that probe message is currently passed through determines, the congestion value is used to indicate the congestion journey in path
Degree；The first leaf equipment receives multiple response messages that the second leaf equipment sends, wherein each sound
Answering message includes the congestion value in the path that the corresponding congestion probe message of the response message is passed through；
The first leaf equipment determines the congestion value per paths according to the multiple response message, from described
At least one path is selected in mulitpath as the path to the second leaf equipment sending data.
2. the method for claim 1, it is characterised in that the congestion value per paths is according to institute
State the congestion value determination of the physical link that path is included.
3. method as claimed in claim 2, it is characterised in that the congestion value of the physical link is root
According to the quantized value and/or the bandwidth profit of correspondence exit port of the queue length of physical link correspondence exit port
Determined with the quantized value of rate.
4. the method as described in claims 1 to 3 any one, it is characterised in that the congestion detection
Message is protocol massages or data message, the congestion probe message also mark including path；It is described to gather around
The mark in plug value and the path is included in frame head, reserved bit or the new field of the congestion probe message
In.
5. the method as described in Claims 1-4 any one, it is characterised in that methods described is also wrapped
Include：
The first leaf equipment receives the reverse blockage probe message that the second leaf equipment sends；From described
The path that the reverse blockage probe message is passed through is obtained in the congestion value field of reverse blockage probe message
Reverse blockage value.
6. the method as described in any one in claim 1-4, it is characterised in that first leaf sets
Standby each response message for receiving also is carried from the second leaf equipment to the path of the first leaf equipment
Congestion value.
7. a kind of leaf equipment based on leaf-ridge topological structure flow load sharing, wherein, the leaf equipment
It is in communication with each other with another leaf equipment, the leaf equipment includes：
Transmitter, for sending multiple congestion probe messages to another leaf equipment by mulitpath,
Each congestion probe message includes congestion value field, and the congestion value field is used to write the congestion detection
The congestion value that the equipment that message is currently passed through determines, the congestion value is used to indicate the Congestion Level SPCC in path；
Receiver, for receiving multiple response messages that another leaf equipment sends, wherein each response
Message includes the congestion value in the path that the corresponding congestion probe message of the response message is passed through；
Processor, for determining the congestion value per paths according to the multiple response message, from described many
At least one path is selected in paths as the path to another leaf equipment sending data.
8. leaf equipment as claimed in claim 7, it is characterised in that the congestion value per paths is basis
What the congestion value of the physical link that the path is included determined.
9. leaf equipment as claimed in claim 8, it is characterised in that the congestion value of the physical link is
The quantized value of the queue length according to physical link correspondence exit port and/or the bandwidth of correspondence exit port
What the quantized value of utilization rate determined.
10. the leaf equipment as described in claim 7 to 9 any one, it is characterised in that the congestion
Probe message is protocol massages or data message, the congestion probe message also mark including path；Institute
The mark for stating congestion value and the path is included in the frame head of the congestion probe message, reserved bit or newly-increased
In field.
11. as described in claim 7 to 10 any one leaf equipment, it is characterised in that the leaf equipment
The path of congestion value minimum in the multiple path is chosen, the data are sent to another leaf equipment.
A kind of 12. ridge equipment based on leaf-ridge topological structure flow load sharing, wherein, the first leaf equipment
It is in communication with each other with the second leaf equipment by the ridge equipment, the ridge equipment includes：
Receiver, for receiving the congestion probe message that the first leaf equipment sends, the congestion detection
The congestion value field of message includes the congestion of the first leaf equipment to the first physical link of the ridge equipment
Value；
Processor, for by the relatively congestion value of first physical link and the ridge equipment to described
The congestion value of the second physical link of the second leaf equipment, determines the first leaf equipment via the ridge equipment
To the congestion value in the path of the second leaf equipment, the congestion value according to the determination processes the congestion and visits
Survey message.
13. ridge equipment as claimed in claim 12, it is characterised in that the processor is specifically for inciting somebody to action
The congestion value of first physical link is defined as institute with the maximum of the congestion value of second physical link
State the congestion value in path.
14. ridge equipment according to claim 13, it is characterised in that when gathering around according to the determination
Plug value process the congestion probe message when, the processor specifically for：When first physical link
Congestion value send the congestion to the second leaf equipment when being the congestion value in the path, directly and detect
Message；When the congestion value that the congestion value of second physical link is the path, the congestion is visited
The congestion value of first physical link surveyed in the congestion value field of message is updated to the second physics chain
The congestion value on road, and send the congestion probe message after the renewal to the second leaf equipment.
15. ridge equipment as claimed in claim 14, it is characterised in that every congestion value of physical link
It is the quantized value and/or the band of correspondence exit port of the queue length that exit port is corresponded to according to the physical link
What the quantized value of utilization rate wide determined.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510981555.3A CN106911584B (en) | 2015-12-23 | 2015-12-23 | Flow load sharing method, device and system based on leaf-ridge topological structure |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510981555.3A CN106911584B (en) | 2015-12-23 | 2015-12-23 | Flow load sharing method, device and system based on leaf-ridge topological structure |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106911584A true CN106911584A (en) | 2017-06-30 |
| CN106911584B CN106911584B (en) | 2020-04-14 |
Family
ID=59200421
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510981555.3A Active CN106911584B (en) | 2015-12-23 | 2015-12-23 | Flow load sharing method, device and system based on leaf-ridge topological structure |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106911584B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108270643A (en) * | 2017-12-14 | 2018-07-10 | 中国银联股份有限公司 | The detection method and equipment of link between Leaf-Spine interchangers |
| CN109728947A (en) * | 2018-12-26 | 2019-05-07 | 成都科来软件有限公司 | A kind of network performance analysis method based on cloud computing in conjunction with network topological diagram |
| CN109802879A (en) * | 2019-01-31 | 2019-05-24 | 新华三技术有限公司 | A kind of flow routing method and device |
| EP3713161A4 (en) * | 2017-11-30 | 2020-12-23 | Huawei Technologies Co., Ltd. | Data transmission method, relevant device and network |
| CN112511325A (en) * | 2019-09-16 | 2021-03-16 | 华为技术有限公司 | Network congestion control method, node, system and storage medium |
| CN112787925A (en) * | 2020-10-12 | 2021-05-11 | 中兴通讯股份有限公司 | Congestion information collection method, optimal path determination method and network switch |
| CN112910795A (en) * | 2021-01-19 | 2021-06-04 | 南京大学 | Edge load balancing method and system based on many sources |
| CN113132225A (en) * | 2019-12-31 | 2021-07-16 | 中兴通讯股份有限公司 | Path selection method and device, computer equipment and computer readable medium |
| CN114221907A (en) * | 2021-12-06 | 2022-03-22 | 北京百度网讯科技有限公司 | Network hash configuration method and device, electronic equipment and storage medium |
| CN114257540A (en) * | 2020-09-10 | 2022-03-29 | 迈络思科技有限公司 | Deadlock free rerouting using detour paths to resolve local link failures |
| CN116192636A (en) * | 2023-04-27 | 2023-05-30 | 苏州浪潮智能科技有限公司 | Network device hash group configuration method and device, electronic device and storage medium |
| CN113132225B (en) * | 2019-12-31 | 2024-04-26 | 中兴通讯股份有限公司 | Path selection method, path selection device, computer equipment and computer readable medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104092628A (en) * | 2014-07-23 | 2014-10-08 | 杭州华三通信技术有限公司 | Flow distribution method and network devices |
| CN104813620A (en) * | 2012-11-20 | 2015-07-29 | 思科技术公司 | Fabric load balancing |
-
2015
- 2015-12-23 CN CN201510981555.3A patent/CN106911584B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104813620A (en) * | 2012-11-20 | 2015-07-29 | 思科技术公司 | Fabric load balancing |
| CN104092628A (en) * | 2014-07-23 | 2014-10-08 | 杭州华三通信技术有限公司 | Flow distribution method and network devices |
Non-Patent Citations (1)
| Title |
|---|
| MOHAMMAD ALIZADEH, TOM EDSALL, SARANG DHARMAPURIKAR ET AL: "CONGA: Distributed Congestion-Aware Load Balancing for Datacenters", 《SIGCOMM 2014，ACM》 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3713161A4 (en) * | 2017-11-30 | 2020-12-23 | Huawei Technologies Co., Ltd. | Data transmission method, relevant device and network |
| CN108270643A (en) * | 2017-12-14 | 2018-07-10 | 中国银联股份有限公司 | The detection method and equipment of link between Leaf-Spine interchangers |
| CN109728947A (en) * | 2018-12-26 | 2019-05-07 | 成都科来软件有限公司 | A kind of network performance analysis method based on cloud computing in conjunction with network topological diagram |
| CN109802879A (en) * | 2019-01-31 | 2019-05-24 | 新华三技术有限公司 | A kind of flow routing method and device |
| CN109802879B (en) * | 2019-01-31 | 2021-05-28 | 新华三技术有限公司 | Data stream routing method and device |
| CN112511325A (en) * | 2019-09-16 | 2021-03-16 | 华为技术有限公司 | Network congestion control method, node, system and storage medium |
| US11888744B2 (en) | 2019-09-16 | 2024-01-30 | Huawei Technologies Co., Ltd. | Spin-leaf network congestion control method, node, system, and storage medium |
| CN112511325B (en) * | 2019-09-16 | 2022-03-11 | 华为技术有限公司 | Network congestion control method, node, system and storage medium |
| CN113132225A (en) * | 2019-12-31 | 2021-07-16 | 中兴通讯股份有限公司 | Path selection method and device, computer equipment and computer readable medium |
| CN113132225B (en) * | 2019-12-31 | 2024-04-26 | 中兴通讯股份有限公司 | Path selection method, path selection device, computer equipment and computer readable medium |
| CN114257540A (en) * | 2020-09-10 | 2022-03-29 | 迈络思科技有限公司 | Deadlock free rerouting using detour paths to resolve local link failures |
| CN114257540B (en) * | 2020-09-10 | 2023-10-27 | 迈络思科技有限公司 | Deadlock free rerouting using detour paths to resolve local link failure |
| CN112787925A (en) * | 2020-10-12 | 2021-05-11 | 中兴通讯股份有限公司 | Congestion information collection method, optimal path determination method and network switch |
| CN112910795A (en) * | 2021-01-19 | 2021-06-04 | 南京大学 | Edge load balancing method and system based on many sources |
| CN112910795B (en) * | 2021-01-19 | 2023-01-06 | 南京大学 | Edge load balancing method and system based on many sources |
| CN114221907A (en) * | 2021-12-06 | 2022-03-22 | 北京百度网讯科技有限公司 | Network hash configuration method and device, electronic equipment and storage medium |
| CN114221907B (en) * | 2021-12-06 | 2023-09-01 | 北京百度网讯科技有限公司 | Network hash configuration method, device, electronic equipment and storage medium |
| CN116192636A (en) * | 2023-04-27 | 2023-05-30 | 苏州浪潮智能科技有限公司 | Network device hash group configuration method and device, electronic device and storage medium |
| CN116192636B (en) * | 2023-04-27 | 2023-08-15 | 苏州浪潮智能科技有限公司 | Network device hash group configuration method and device, electronic device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106911584B (en) | 2020-04-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106911584A (en) | It is a kind of based on leaf-flow load sharing method of ridge topological structure, apparatus and system | |
| CN112738820B (en) | Dynamic deployment method and device of service function chain and computer equipment | |
| CN104852857B (en) | Distributed data transport method and system based on load balancing | |
| US10110467B2 (en) | Routing path selection based on dynamical metric values | |
| CN102474454B (en) | Inter-node link aggregation system and method | |
| US20170339039A1 (en) | Method for devices in a network to participate in an end-to-end measurement of latency | |
| US10355971B2 (en) | System and method for data path validation and verification | |
| CN104767682A (en) | Routing method and system as well as routing information distributing method and device | |
| CN103929782B (en) | A kind of resources balance multi-path route method suitable for industrial wireless sensor network | |
| JP2002199009A (en) | Method for measuring available band by variable speed probing using zoom in/out method | |
| CN105556916B (en) | The information statistical method and device of network flow | |
| CN109639575A (en) | Route planning method based on link congestion coefficient | |
| US20160065449A1 (en) | Bandwidth-Weighted Equal Cost Multi-Path Routing | |
| CN104917659B (en) | A kind of mapping method of virtual network based on virtual network connection performance | |
| CN105634968B (en) | Device and method for the transmission controlled the data flows | |
| CN109617805B (en) | Method and device for acquiring link dynamic attribute and method and device for selecting path | |
| CN108234320A (en) | Message transmitting method and interchanger | |
| CN106713378A (en) | Method and system for realizing service provision by multiple application servers | |
| CN106656679A (en) | Available bandwidth measurement method and device | |
| CN105262534B (en) | A kind of method for routing and device suitable for satellite communication network | |
| CN109889444A (en) | A kind of methods, devices and systems of planning path | |
| CN114915995A (en) | Network slice monitoring method, system and storage medium based on in-band network telemetry | |
| US8107364B2 (en) | Communication apparatus | |
| CN105763468B (en) | A kind of transmission method and device of bgp update message | |
| US11190430B2 (en) | Determining the bandwidth of a communication link |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |