CN107046507A - It is a kind of to be used for the jamming control method of multiplexed transport in DCN - Google Patents
It is a kind of to be used for the jamming control method of multiplexed transport in DCN Download PDFInfo
- Publication number
- CN107046507A CN107046507A CN201611147795.4A CN201611147795A CN107046507A CN 107046507 A CN107046507 A CN 107046507A CN 201611147795 A CN201611147795 A CN 201611147795A CN 107046507 A CN107046507 A CN 107046507A
- Authority
- CN
- China
- Prior art keywords
- task
- data
- dcn
- current
- cwnd
- 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
- H04L47/27—Evaluation or update of window size, e.g. using information derived from acknowledged [ACK] packets
-
- 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
- H04L47/29—Flow control; Congestion control using a combination of thresholds
-
- 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
- H04L47/37—Slow start
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
It is used for the jamming control method of multiplexed transport in DCN the invention discloses a kind of, the data volume sent according to current task and stream accelerates the transmission rate of hangover stream in small task and task.Task conditions of streaking caused by the choking phenomenon caused present invention, avoiding the big task in DCN networks and a small number of streams at a slow speed, efficiently reduces the average completion time of task, lifts the overall performance of network application.
Description
Technical field
Present invention design is a kind of to be used for gathering around for multiplexed transport in DCN (data center network, Data Center Network)
Fill in control method.
Background technology
The appearance of data center network calculating platform is brought to online service (such as web search, online retail, advertisement)
Huge income.These data-intensive services have the characteristics that：Each inquiry can be seen as a task, Mei Gecha
Inquiry is distributed to thousands of leaf node, and last leaf node completes inquiry and returns result to distribution node, only owns
Stream, which all completes task, to be completed at last, i.e. the completion of task needs every stream all to complete.
As the Internet, applications become high diversity and complication, in order to solve the deficiency of Traditional control host-host protocol,
Improve the deadline of task rank, domestic and international many scholars have done many researchs for the deadline of reduction task rank, seek
Seek some practicable method reduction task completion times.
DCTCP (Data Center TCP) agreements are proposed aiming at the data center network of high bandwidth, the application of high fan-in
Replacement conventional TCP protocols improvement host-host protocol.It utilizes ecn (explicit congestion notification) (Explicit Congestion
Notification, ECN) judge congestion state.DCTCP discusses afterbody delay comprehensively for the first time, is being changed without what is deposited at a low speed
On the premise of commercial switches, DCN (data center network) high burst tolerance, low latency, high-throughput is realized.But,
DCTCP problems are that it does not have task perceptibility.
L2DCT (Low Latency Data Center Transport) reduces short stream by improving the priority of short stream
Deadline, and then reduce the mean flow deadline.In congestion avoidance phase, the congestion of short stream within each bag two-way time
Window increase by 1, and growing stream only increases by 0.05.When detecting congestion, it is many that the congestion window of long stream declines, and short stream is gathered around
What plug window declined lacks.L2DCT does well in the deadline for reducing short stream, but compromises the handling capacity of long stream, simultaneously
Also it is not based on task rank optimization performance.
Baraate be one do not need network in respective switch cooperated too much distributed task scheduling sensitivity scheduling
System, the purpose of design is the distributed scheduling scheme for realizing FIFO-LM.Baraate is that each task distributes an overall situation
Unique task id, the task that regulation possesses smaller No. id has higher priority, and arriving first for task distributes smaller id
Number, that is, realize the priority control of first in first out.Baraate can reduce average task completion time, but deployable compared with
Difference.
Varys agreements are intended to reduce the call duration time of task in data-intensive data center network.Varys is a number
According to intensive framework, keep high network utilization and ensure that task will not be hungry.The scheduling strategy that Varys has been used for reference
It is that, based on coflow, but Varys is unintelligible on coflow division, it is realized and deployment has difficulties.
It is therefore proposed that effectively reducing average task completion time and the data center network transport layer association with deployable
View, becomes a urgent problem to be solved.
The content of the invention
The technical problems to be solved by the invention are that being used for task in DCN there is provided one kind in view of the shortcomings of the prior art passes
Defeated jamming control method, this method can effectively reduce the average completion time of data task in data center network.
The technical solution of invention is as follows：
It is a kind of to be used for the jamming control method of multiplexed transport in DCN, comprise the following steps：
Step one：Parameter initialization；Initialization sends window size cwnd, sends congestion flag position (the i.e. CE of packet
Position)；Small task data amount threshold value taskmin, big task data amount threshold value taskmax；
Step 2：It is recipient or sender to judge work at present end, if recipient turns to step 8 and otherwise turned
To step 3；
Step 3：Sender receives ACK bags (confirming bag)；After current send window intraoral whole ACK bags are received, sentence
The congestion flag position for whether having ACK bags in disconnected whole ACK bags is set to 1, if the congestion flag position of ACK bags is set to 1, then
The quantity that 1 ACK bags are set to according to congestion flag position calculates the Congestion Level SPCC α of current data packet round trip cyclenAnd turn to step
Rapid four；Otherwise judge whether to be in slow start stage, cwnd is updated if in slow start stage for 2 × cwnd, is otherwise updated
Cwnd is cwnd+1, and turns to step 7；
Step 4：Sender is according to the current task data accepted amount S returned in ACK bag added fieldstAnd it is large and small
Task data amount threshold calculations mission factor β, and turn to step 5；
Step 5：Sender updates the data volume S for currently flowing and having sentf, with reference to current task data accepted amount StWith work as
The number n flowed in preceding task calculates stream tailing factor γ, and turns to step 6；
Step 6：Sender is according to Step 3: the α that four, five calculating are obtainedn, β, γ update and send window size cwnd simultaneously
Turn to step 7；
Step 7：Sender sends cwnd packet in packet round-trip delay, and judges whether data have sent
Finish, otherwise return to step three if not being sent terminate data transfer；
Step 8：Recipient's received data packet simultaneously checks whether the congestion flag position of packet header is set to 1, if
It is then 1 by the congestion flag position of its corresponding ACK bag, is otherwise set to 0, turns to step 9；The wherein congestion of packet header
Flag bit is configured by interchanger, exchanges machine monitoring current queue size, if queue length exceedes given threshold, immediately will
The congestion flag position for being currently entering the packet of queue is 1；
Step 9：Recipient updates current task data accepted amount StWith the number n flowed in current task, and be stored in pair
Answer the added field in ACK bags；Then ACK bags are sent to sender, and turn to step 8.
The Congestion Level SPCC α of current data packet round trip cycle in the step 3nCalculation formula be：
Wherein, m be in the intraoral all ACK bags of current send window congestion flag position be set to 1 ACK bags quantity, αn-1Table
Show the Congestion Level SPCC of a packet round trip cycle, g represents moving average weights.
The moving average weights g is set to 0.625.
Mission factor β calculation formula is in the step 4：
Wherein, StFor current task data accepted amount, taskminFor small task data amount threshold value, taskmaxFor big task
Data-quantity threshold.
Tailing factor γ calculation formula is in the step 5：
Wherein, SfCurrently to flow the data volume sent, StFor current task data accepted amount, (i.e. task has sent number
According to amount), n is the number that flows in current task.
In the step 6：Sender is according to Step 3: the α that four, five calculating are obtainedn, β, γ update send window size
Cwnd formula is：
The formula can not only adjust the size for sending window according to Congestion Level SPCC, can also accelerate the transmission speed of small task
Rate, it is to avoid the choking phenomenon that big multiplexed transport is caused, while can speed up the transmission rate of hangover stream inside task, is reduced flat
Equal task completion time.
Initialization includes transmission window size cwnd being set to 2 in the step one, by the congestion of transmitted packet
Flag bit is set to 0, by small task data amount threshold value taskminIt is set to 100KB, big task data amount threshold value taskmaxSet
For 1MB.
Beneficial effect：
The data volume optimization task average completion time that the present invention has been sent according to current task and stream.Occurring congestion
When, by carrying out different punitive measures according to the difference for having sent size to different tasks, hangover is flowed inside task
Take lighter punitive measures, accelerate the transmission rate of hangover stream in small task and task, it is to avoid the big task in DCN networks
Task conditions of streaking caused by the choking phenomenon caused and a small number of streams at a slow speed, efficiently reduces the average completion time of task,
Lift the overall performance of network application.
Actual measurement shows that this method can effectively improve average task completion time, detailed in Example.
Brief description of the drawings
Fig. 1 is flow chart of the invention.
Fig. 2 is the topological schematic diagram of a simple data center network.
Fig. 3 is the inventive method (TL2DCT), (stream size is constant inside task, changes for DCTCP, L2DCT change task size
The number n) flowed in change task average task completion time.
It is complete that Fig. 4 is the inventive method (TL2DCT), DCTCP, L2DCT change the average task of stream time started inside task
Into the time.
Fig. 5 is that the inventive method (TL2DCT), DCTCP, L2DCT normalized average task under extensive environment are completed
Time；Fig. 5 (a) is different quantile task completion times, and Fig. 5 (b) is the average task completion time of different type task.
Embodiment
The present invention is described in further details below with reference to the drawings and specific embodiments：
Embodiment 1：
Referring to Fig. 1, Fig. 1 is flow chart of the invention：A kind of transmission controlling party for being used to feed back based on multilevel congestion in DCN
Method, process is as follows：
The initial window cwnd that sends is dimensioned to 2 by sender, and queue length threshold is set to 65 by interchanger, will be gathered around
Plug flag bit is CE and is set to 0, by small task data amount threshold value taskminIt is set to 100KB, big task data amount threshold value
taskmaxIt is set to 1MB；
When exchanging machine monitoring current queue size more than threshold value, the congestion flag of the packet of queue will be currently entering immediately
Position is 1, and after the packet that congestion flag position is set to 1 reaches recipient, recipient is by the corresponding ACK bags of the packet
Congestion flag position is 1, and ACK bags are sent into sender.
Sender is received after the intraoral whole ACK bags of current send window, and 1 ACK bags are set to according to congestion flag position
Quantity calculate current data packet round trip cycle Congestion Level SPCC αn：
Wherein, m is set to the quantity of 1 ACK bags, α for CE in the intraoral all ACK bags of current send windown-1Represent upper one
The Congestion Level SPCC of individual packet round trip cycle, g represents moving average weights, is set to 0.625.
The current task data accepted amount S that sender feeds back according to ACK bagst, according to following formula calculating task
Factor-beta：
Sender's statistics currently flows the data volume S sentf, number has been received with reference to the current task that ACK bags are fed back
According to amount StWith the number n flowed in current task, tailing factor γ is calculated：
Recipient passes through the α that calculates beforen, β, γ update current window size cwnd：
The cwnd that window cwnd is updated to newly calculate will be sent.
For the effect of the checking present invention, performance test has been carried out on NS2.35 network simulation platforms.
Fig. 2 illustrates the many-to-one topology for occurring to pass through single bottleneck in data center network.Data-center applications
In, in order to ensure the high scalability and reliability of service, application data is typically that cutting is stored in different servers, each service
The data slot of device storage is referred to as server request unit (Server Request Unit, SRU).Generally, in data center
Portion is communicated in such a way：Convergence server sends request of data to all working server.Each workspace server is received
To after request of data, stored data slot, i.e. SRU will be transmitted.Convergence server is received after the SRU of all requests, will
Data slot merges or handled, and then sends next round request of data.
Fig. 3 compares average of DCTCP, L2DCT and the inventive method (TL2DCT) under different task size cases
It is engaged in the deadline.Wherein, 100 main frames each send 1 stream to a convergence main frame simultaneously, and every stream size is 100KB.It is real
Test comprising three groups of contrast experiments, be that task sets different flow amount n (task width), to change the size of task.First group
In experiment, there are 20 small tasks, each small task includes two streams；Separately there is 20 big task, each task is included
Three streams.In second group of experiment, there are 20 small tasks, each small task includes two streams；Separately there is 15 big task,
Each big task includes four streams.In 3rd group of experiment, there are 20 small tasks, each small task includes two streams；Separately deposit
In 12 big task, each task includes five streams.It can be seen from figure 3 that because TL2DCT accelerates the transmission rate of small task,
The choking phenomenon that big multiplexed transport is caused is avoided, average task completion time is eventually reduced.Wherein, with size task
Stand out away from increase, TL2DCT income is more obvious.For example, in the 3rd group of experiment, L2DCT average task completion time phase
Than reducing 15% and 11.9% respectively in DCTCP and L2DCT.
Fig. 4 shows the average task completion time for occurring hangover stream inside DCTCP, L2DCT and TL2DCT task.Its
In, 100 main frames each send 1 stream to a convergence main frame simultaneously, and every stream size is 100KB.There are 50 in experiment
Task, each task includes two streams.1st article of stream starts simultaneously at transmission data, during the interval that the 2nd article of stream is sent away from the 1st article of stream
Between be uniformly distributed respectively in 0-10ms, 0-20ms, 0-30ms, 0-40ms.Fig. 3 is shown, compared to DCTCP and L2DCT,
TL2DCT average task completion time is greatly reduced.Interval is increased over time, and the hair of hangover stream can be substantially improved in TL2DCT
Transmission rate, further speeds up task and completes speed.
Fig. 5 shows acceleration effects of the TL2DCT under True Data central site network discharge model.Wherein, 100 main frames
2038 streams are sent by same interchanger altogether to a convergence main frame.These streams constitute four generic tasks, specifically include narrow short
Task, long narrow task, wide short task and wide long task.Wherein, the width of 52 narrow short task is 3, and stream size is all 10KB；16
The width of individual long narrow task is 7, and stream size is all 200KB；The width of 15 wide short task is 50, and stream size is all 9KB；17
The width 60 of wide long task, stream size is all 3MB.The transmission time started of task is obeyed inside poisson arrival distribution, task
The transmission time of all streams is evenly distributed in 20ms.Fig. 5 (a) experimental results show, percent 50 before TL2DCT, 75,95,
The average task completion time of 100 completing for tasks reduces 100% compared to DCTCP respectively, 36.36%, 17.8%,
14%, reduce 80%, 31.8%, 24%, 13.9% compared to L2DCT.Fig. 5 (b) experimental results are shown, narrow short in TL2DCT
The average task completion time of task and long narrow task reduces 98% and 53% respectively compared to DCTCP, subtracts compared to L2DCT
80% and 46% are lacked, this explanation TL2DCT can efficiently reduce the deadline of narrow task according to task width.Due to narrow
68%, the TL2DCT that business accounts for general assignment number reduces overall average task completion time.
Claims (7)
1. a kind of be used for the jamming control method of multiplexed transport in DCN, it is characterised in that comprises the following steps：
Step one：Parameter initialization；Initialization sends window size cwnd, sends the congestion flag position of packet；Small number of tasks
According to amount threshold value taskmin, big task data amount threshold value taskmax；
Step 2：It is recipient or sender to judge work at present end, if recipient turns to step 8 and otherwise turns to step
Rapid three；
Step 3：Sender receives ACK bags；After current send window intraoral whole ACK bags are received, whole ACK bags are judged
In whether have congestion flag position be set to 1 ACK bags, if the congestion flag of ACK bags is set to 1, then foundation congestion mark
Will position is set to the Congestion Level SPCC α of the quantity calculating current data packet round trip cycle of 1 ACK bagsnAnd turn to step 4；Otherwise sentence
It is disconnected whether to be in slow start stage, cwnd is updated if in slow start stage for 2 × cwnd, it is cwnd+ otherwise to update cwnd
1, and turn to step 7；
Step 4：Sender is according to the current task data accepted amount S returned in ACK bag added fieldstAnd large and small task
Data-quantity threshold calculating task factor-beta, and turn to step 5；
Step 5：Sender updates the data volume S for currently flowing and having sentf, with reference to current task data accepted amount StIt is former with working as
The number n flowed in business calculates stream tailing factor γ, and turns to step 6；
Step 6：Sender is according to Step 3: the α that four, five calculating are obtainedn, β and γ update and send window size cwnd and turn to
Step 7；
Step 7：Sender sends cwnd packet in packet round-trip delay, and judges whether data are sent, such as
Fruit is not sent then return to step three, otherwise terminates data transfer；
Step 8：Recipient's received data packet simultaneously checks whether the congestion flag position of packet header is set to 1, if it is will
Congestion flag position in its corresponding ACK bag is 1, is otherwise set to 0, turns to step 9；
Step 9：Recipient updates current task data accepted amount StWith the number n flowed in current task, and corresponding A CK is stored in
Added field in bag；Then ACK bags are sent to sender, and turn to step 8.
2. according to claim 1 be used for the jamming control method of multiplexed transport in DCN, it is characterised in that the step
The Congestion Level SPCC α of current data packet round trip cycle in threenCalculation formula be：
Wherein, m be in the intraoral all ACK bags of current send window congestion flag position be set to 1 ACK bags quantity, αn-1In expression
The Congestion Level SPCC of one packet round trip cycle, g represents moving average weights.
3. according to claim 2 be used for the jamming control method of multiplexed transport in DCN, it is characterised in that the slip
Average weight g is set to 0.625.
4. according to claim 2 be used for the jamming control method of multiplexed transport in DCN, it is characterised in that the step
Mission factor β calculation formula is in four：
Wherein, StFor current task data accepted amount, taskminFor small task data amount threshold value, taskmaxFor big task data
Measure threshold value.
5. according to claim 4 be used for the jamming control method of multiplexed transport in DCN, it is characterised in that the step
Tailing factor γ calculation formula is in five：
Wherein, SfCurrently to flow the data volume sent, StFor current task data accepted amount, n is the number flowed in current task
Mesh.
6. according to claim 5 be used for the jamming control method of multiplexed transport in DCN, it is characterised in that the step
In six：Sender is according to Step 3: the α that four, five calculating are obtainedn, β, γ update and send window size cwnd formula and be：
7. being used for the jamming control method of multiplexed transport in DCN according to according to any one of claims 1 to 6, its feature exists
In initialization includes in the step one：Window size cwnd will be sent and be set to 2, by the congestion flag of transmitted packet
Position is set to 0, by small task data amount threshold value taskminIt is set to 100KB, big task data amount threshold value taskmaxIt is set to
1MB。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611147795.4A CN107046507B (en) | 2016-12-13 | 2016-12-13 | A kind of jamming control method for multiplexed transport in DCN |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611147795.4A CN107046507B (en) | 2016-12-13 | 2016-12-13 | A kind of jamming control method for multiplexed transport in DCN |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107046507A true CN107046507A (en) | 2017-08-15 |
| CN107046507B CN107046507B (en) | 2019-05-28 |
Family
ID=59543839
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611147795.4A Active CN107046507B (en) | 2016-12-13 | 2016-12-13 | A kind of jamming control method for multiplexed transport in DCN |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107046507B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107770082A (en) * | 2017-10-19 | 2018-03-06 | 中南大学 | A kind of transfer control method of task based access control discharge characteristic in data center network |
| CN110061927A (en) * | 2019-04-26 | 2019-07-26 | 东南大学 | Congestion aware and labeling method towards micro- burst flow in a kind of more queuing data center environments |
| CN110351187A (en) * | 2019-08-02 | 2019-10-18 | 中南大学 | Data center network Road diameter switches the adaptive load-balancing method of granularity |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130044598A1 (en) * | 2011-08-19 | 2013-02-21 | Futurewei Technologies, Inc. | System and Method for Transmission Control Protocol Slow-Start |
| CN104702527A (en) * | 2015-03-24 | 2015-06-10 | 大连理工大学 | Method for controlling multi-priority connection congestion time windows in multi-path TCP |
| CN105827540A (en) * | 2016-05-30 | 2016-08-03 | 清华大学深圳研究生院 | Data center network transmission layer stream data transmission method based on priority |
| CN106059951A (en) * | 2016-06-08 | 2016-10-26 | 中南大学 | Transmission control method for DCN (Data Center Network) based on multilevel congestion feedback |
-
2016
- 2016-12-13 CN CN201611147795.4A patent/CN107046507B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130044598A1 (en) * | 2011-08-19 | 2013-02-21 | Futurewei Technologies, Inc. | System and Method for Transmission Control Protocol Slow-Start |
| CN104702527A (en) * | 2015-03-24 | 2015-06-10 | 大连理工大学 | Method for controlling multi-priority connection congestion time windows in multi-path TCP |
| CN105827540A (en) * | 2016-05-30 | 2016-08-03 | 清华大学深圳研究生院 | Data center network transmission layer stream data transmission method based on priority |
| CN106059951A (en) * | 2016-06-08 | 2016-10-26 | 中南大学 | Transmission control method for DCN (Data Center Network) based on multilevel congestion feedback |
Non-Patent Citations (1)
| Title |
|---|
| 马矿，何加铭等，: ""基于MPTCP的自适应拥塞控制算法"", 《移动通信》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107770082A (en) * | 2017-10-19 | 2018-03-06 | 中南大学 | A kind of transfer control method of task based access control discharge characteristic in data center network |
| CN107770082B (en) * | 2017-10-19 | 2020-05-12 | 中南大学 | Transmission control method based on task flow characteristics in data center network |
| CN110061927A (en) * | 2019-04-26 | 2019-07-26 | 东南大学 | Congestion aware and labeling method towards micro- burst flow in a kind of more queuing data center environments |
| CN110061927B (en) * | 2019-04-26 | 2021-07-23 | 东南大学 | Congestion sensing and marking method facing micro burst flow in multi-queue data center environment |
| CN110351187A (en) * | 2019-08-02 | 2019-10-18 | 中南大学 | Data center network Road diameter switches the adaptive load-balancing method of granularity |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107046507B (en) | 2019-05-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105933232B (en) | Support the Multipath Transmission control terminal and method of multi-service data transmission demand | |
| CN106059951B (en) | A kind of transfer control method for based on multilevel congestion feedback in DCN | |
| CN106302228B (en) | The transfer control method of task based access control perception in a kind of data center network | |
| CN101562566B (en) | Active queue management method based on real-time router cache occupancy rate | |
| CN104009931B (en) | The jamming control method cooperateed with data center network based on advertised window and multithread | |
| CN107948103A (en) | A kind of interchanger PFC control methods and control system based on prediction | |
| CN103647722B (en) | A kind of Link Congestion Control Method based on prestige | |
| CN107046507A (en) | It is a kind of to be used for the jamming control method of multiplexed transport in DCN | |
| CN105490962B (en) | A kind of QoS management methods based on OpenFlow networks | |
| CN107566275B (en) | Multi-path transmission method based on the delay inequality opposite sex in data center network | |
| Zou et al. | Flow-aware adaptive pacing to mitigate TCP incast in data center networks | |
| CN107396396A (en) | Support the data transfer management method of multi-source multipath | |
| CN106302227A (en) | Hybrid network stream scheduling method and switch | |
| CN106559330A (en) | A kind of dynamic path planning method based on SDN | |
| CN109714267A (en) | Manage the transfer control method and system of reversed queue | |
| CN108540395A (en) | Without the congestion judgment method lost in network | |
| CN106027407A (en) | Adjusting method of congestion windows in network, network congestion control method and interchanger | |
| CN107154897A (en) | Isomery stream partition method based on bag scattering in DCN | |
| CN109257282A (en) | A kind of data transmission method and device | |
| CN107800643A (en) | Message forwarding method and device | |
| CN103428104A (en) | Content-center network-based congestion control method | |
| CN102752076B (en) | Control method that data send and device and computer system | |
| CN103152649B (en) | An a kind of streaming media distribution transmission point rank subtracts frame control method automatically | |
| CN107770082A (en) | A kind of transfer control method of task based access control discharge characteristic in data center network | |
| CN104869113A (en) | Network transmission method and system based on minimum completion time |
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 |