CN106873915A - A kind of data transmission method and device based on RDMA registers memory blocks - Google Patents
A kind of data transmission method and device based on RDMA registers memory blocks Download PDFInfo
- Publication number
- CN106873915A CN106873915A CN201710096131.8A CN201710096131A CN106873915A CN 106873915 A CN106873915 A CN 106873915A CN 201710096131 A CN201710096131 A CN 201710096131A CN 106873915 A CN106873915 A CN 106873915A
- Authority
- CN
- China
- Prior art keywords
- memory
- memory block
- registered
- rdma
- transmitting terminal
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0602—Interfaces specially adapted for storage systems specifically adapted to achieve a particular effect
- G06F3/061—Improving I/O performance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0602—Interfaces specially adapted for storage systems specifically adapted to achieve a particular effect
- G06F3/0614—Improving the reliability of storage systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0628—Interfaces specially adapted for storage systems making use of a particular technique
- G06F3/0638—Organizing or formatting or addressing of data
- G06F3/064—Management of blocks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0628—Interfaces specially adapted for storage systems making use of a particular technique
- G06F3/0646—Horizontal data movement in storage systems, i.e. moving data in between storage devices or systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Bus Control (AREA)
- Communication Control (AREA)
Abstract
This application discloses a kind of data transmission method and device based on RDMA registers memory blocks, the method includes application bulk memory, and the bulk memory is divided into multiple memory blocks for having and presetting size；The memory block is registered to the transmitting terminal and receiving terminal of network interface card, wherein the number for being registered to the first memory block of receiving terminal is not less than the number of the second memory block for being registered to transmitting terminal；Available second memory block is obtained, and loads data to be sent；Using network interface card by data is activation to be sent to first memory block.Above-mentioned data transmission method and device based on RDMA registers memory blocks, can reduce occupancy of the RDMA to memory source, reduce the system resource that RDMA takes because calculating operation registers memory, so as to improve the stability of data transfer, and improve data transmission efficiency.
Description
Technical field
The invention belongs to high performance communication technical field, more particularly to a kind of data based on RDMA registers memory blocks are passed
Transmission method and device.
Background technology
At present, common network interface card is integrated with the function of supporting hardware check sum, and software is improved, so as to reduce
The copy number of data is sent, but the copy number for receiving data cannot be reduced, and this copied part amount will take a large amount of calculating of CPU
Cycle.The course of work of common network interface card is as follows：First on the data pack buffer for receiving to system, packet after treatment, phase
Answer data to be assigned to a TCP to connect；Then, reception system is again unsolicited tcp data with corresponding application program
Connect, and copy data to target storage address from system buffer.So, the factor of restriction network rate is occurred as soon as
：Application communication intensity be continuously increased and host CPU the task of processing data is heavy between kernel and application memory makes system
Host CPU resources are constantly added, efficient software and strengthening system load management is configured, the key of problem is main frame to be eliminated
Message delay between frequent data item transmission unnecessary in CPU, reduction system.
RDMA is the direct incoming computer storage area registered in advance of data, by data a from system by network
Move quickly into remote system stored device, without being had any impact to operating system, how much counted so there is no need to use
The processing function of calculation machine.It eliminates external memory storage and replicates and text exchange operation, thus can vacate bus space and CPU weeks
Phase is used to improve application system performance.And at present general way need to by system first incoming information is analyzed with mark,
Then it is then stored into correct region.Realization using RDMA technologies in ceph clusters typically has two kinds of selections：One is base
In the realization of Open Source Code accelio class libraries, but due to accelio class bank code complexities are higher, and accelio class libraries sheets
Body there is a problem of certain, in the case where these problems are not solved, be not suitable for being applied to production environment；The second is being based on
VPI verbs interfaces are self-built, realize safety, controllable bottom communication framework.Realize that RDMA technologies exist using the second way
Application in ceph clusters, is major way of the current RDMA technologies in field of storage application, and one of its core is to design base
In RDMA efficiently, stabilization, controllable registers memory block administrative mechanism.
In order that RDMA technologies are applied in distributed ceph clusters answered for upper strata, it is necessary to encapsulate a set of communication interface
With calling, Open Source Code accelio class libraries realizes the traffic model based on RDMA technologies, and realizes the XIO of application layer and connect
Mouthful, but there are some stability in accelio class libraries, can not realize commercialization in distributed ceph clusters in itself
Using.In order to realize commercialization of the RDMA technologies in distributed ceph clusters, it may be considered that based on the self-built realization of Verbs interfaces
High-performance bottom communication module based on RDMA, and one of self-built core be design it is a kind of efficiently, stabilization, controllable registration
Memory block management mechanism.
The content of the invention
To solve the above problems, the invention provides a kind of data transmission method and dress based on RDMA registers memory blocks
Put, occupancy of the RDMA to memory source can be reduced, reduce the system resource that RDMA takes because calculating operation registers memory, from
And the stability of data transfer is improved, and improve data transmission efficiency.
A kind of data transmission method based on RDMA registers memory blocks that the present invention is provided, including：
Application bulk memory, and the bulk memory is divided into multiple memory blocks for having and presetting size；
The memory block is registered to the transmitting terminal and receiving terminal of network interface card, wherein being registered to the first memory block of receiving terminal
Number is not less than the number of the second memory block for being registered to transmitting terminal；
Available second memory block is obtained, and loads data to be sent；
Using network interface card by data is activation to be sent to first memory block.
Preferably, in the above-mentioned data transmission method based on RDMA registers memory blocks,
It is described the memory block is registered to network interface card transmitting terminal and receiving terminal be：
Memory block is registered to transmitting terminal and the reception of Infiniband network interface cards using VPI verbs interfaces ibv_reg_mr
End.
Preferably, in the above-mentioned data transmission method based on RDMA registers memory blocks,
It is described the bulk memory is divided into multiple memory blocks with default size to be：
The bulk memory is divided into the memory block that multiple sizes are 8K.
Preferably, in the above-mentioned data transmission method based on RDMA registers memory blocks,
The number of first memory block for being registered to receiving terminal is individual not less than the second memory block for being registered to transmitting terminal
Number is：
The number for being registered to the first memory block of receiving terminal is more than the number of the second memory block for being registered to transmitting terminal.
Preferably, in the above-mentioned data transmission method based on RDMA registers memory blocks,
After the utilization network interface card is by data is activation to be sent to first memory block, also include：
The receiving terminal notifies that the transmitting terminal data transfer is completed, and the memory block that transmitting terminal takes is set into available shape
State.
A kind of data transmission device based on RDMA registers memory blocks that the present invention is provided, including：
Blocking unit, for applying for bulk memory, and is divided into multiple internal memories for having and presetting size by the bulk memory
Block；
Registering unit, transmitting terminal and receiving terminal for the memory block to be registered to network interface card, wherein being registered to receiving terminal
The first memory block number be not less than be registered to transmitting terminal the second memory block number；
Loading location, for obtaining available second memory block, and loads data to be sent；
Transmitting element, for utilizing network interface card by data is activation to be sent to first memory block.
Preferably, in the above-mentioned data transmission device based on RDMA registers memory blocks,
Specifically for be registered to for memory block using VPI verbs interfaces ibv_reg_mr by the registering unit
The transmitting terminal and receiving terminal of Infiniband network interface cards.
Preferably, in the above-mentioned data transmission device based on RDMA registers memory blocks,
The blocking unit by the bulk memory specifically for being divided into the memory block that multiple sizes are 8K.
Preferably, in the above-mentioned data transmission device based on RDMA registers memory blocks,
The registering unit is registered to reception specifically for the memory block to be registered to the transmitting terminal and receiving terminal of network interface card
The number of first memory block at end is more than the number of the second memory block for being registered to transmitting terminal.
Preferably, in the above-mentioned data transmission device based on RDMA registers memory blocks, also include：
Notification unit, for notifying that the transmitting terminal data transfer is completed using the receiving terminal, and transmitting terminal is taken
Memory block be set to upstate.
The above-mentioned data transmission method and dress based on RDMA registers memory blocks provided by foregoing description, the present invention
Put, because the method includes：Application bulk memory, and the bulk memory is divided into multiple memory blocks for having and presetting size；
The memory block is registered to the transmitting terminal and receiving terminal of network interface card, wherein the number for being registered to the first memory block of receiving terminal is not small
In the number of the second memory block for being registered to transmitting terminal；Available second memory block is obtained, and loads data to be sent；Using net
Card, therefore, it is possible to reduce occupancy of the RDMA to memory source, is reduced in data is activation to be sent to first memory block
The system resource that RDMA takes because calculating operation registers memory, so as to improve the stability of data transfer, and improves data biography
Defeated efficiency.
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 to embodiment or existing
The accompanying drawing to be used needed for having technology description is briefly described, it should be apparent that, drawings in the following description are only this
Inventive embodiment, for those of ordinary skill in the art, on the premise of not paying creative work, can also basis
The accompanying drawing of offer obtains other accompanying drawings.
The signal of the first data transmission method based on RDMA registers memory blocks that Fig. 1 is provided for the embodiment of the present application
Figure；
Fig. 2 is the principle schematic of the administrative mechanism of communication memory block；
The signal of the first data transmission device based on RDMA registers memory blocks that Fig. 3 is provided for the embodiment of the present application
Figure.
Specific embodiment
Core concept of the invention is to provide a kind of data transmission method and device based on RDMA registers memory blocks, energy
Occupancy of the RDMA to memory source is enough reduced, the system resource that RDMA takes because calculating operation registers memory is reduced, so as to carry
The stability of data transfer high, and improve data transmission efficiency.
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation is described, it is clear that described embodiment is only a part of embodiment of the invention, rather than whole embodiments.It is based on
Embodiment in the present invention, it is every other that those of ordinary skill in the art are obtained under the premise of creative work is not made
Embodiment, belongs to the scope of protection of the invention.
The first data transmission method based on RDMA registers memory blocks that the embodiment of the present application is provided is as shown in figure 1, Fig. 1
It is the schematic diagram of the first data transmission method based on RDMA registers memory blocks that the embodiment of the present application is provided, the method includes
Following steps：
S1：Application bulk memory, and the bulk memory is divided into multiple memory blocks for having and presetting size；
It should be noted that the bulk memory that can apply is dispersed as the memory block (chunk) of 8K mono-.
S2：The memory block is registered to the transmitting terminal and receiving terminal of network interface card, wherein being registered to the first internal memory of receiving terminal
The number of block is not less than the number of the second memory block for being registered to transmitting terminal；
It should be noted that the network interface card can be preferably the network interface card with Infiniband frameworks, RDMA is to realize
A kind of popular protocol of Infiniband frameworks.
S3：Available second memory block is obtained, and loads data to be sent；
S4：Using network interface card by data is activation to be sent to first memory block.
Specifically, with reference to Fig. 2, Fig. 2 be communicate memory block administrative mechanism principle schematic, before sending data first from
Enough memory blocks are obtained in free memory pools, when data are written to end memory, the memory block that will be got is put back into
Free memory pools.If free memory pools do not have enough memory blocks, it then follows how much is how many memory block transmission, remainder
Retransmited until having memory block in free memory pools；If do not have memory block in free memory pools, message is blocked
Retransmited after having memory block in free memory pools.
By foregoing description, above-mentioned the first data based on RDMA registers memory blocks that the embodiment of the present application is provided
Transmission method, due to including application bulk memory, and is divided into multiple memory blocks with default size by the bulk memory；Will
The memory block is registered to the transmitting terminal and receiving terminal of network interface card, wherein the number for being registered to the first memory block of receiving terminal is not less than
It is registered to the number of the second memory block of transmitting terminal；Available second memory block is obtained, and loads data to be sent；Using network interface card
By in data is activation to be sent to first memory block, therefore, it is possible to reduce occupancy of the RDMA to memory source, RDMA is reduced
The system resource taken because calculating operation registers memory, so as to improve the stability of data transfer, and improves data transfer effect
Rate.
Second data transmission method based on RDMA registers memory blocks that the embodiment of the present application is provided, is above-mentioned first
Plant and be based on the basis of the data transmission method of RDMA registers memory blocks, also including following technical characteristic：
It is described the memory block is registered to network interface card transmitting terminal and receiving terminal be：
Memory block is registered to transmitting terminal and the reception of Infiniband network interface cards using VPI verbs interfaces ibv_reg_mr
End.
It should be noted that in RDMA communication process, the memory block registered by these is led to the content of transmitting terminal
Cross Infiniband network interface cards and be transferred to receiving terminal, the pressure that data transfer is caused to CPU can be reduced.
The third data transmission method based on RDMA registers memory blocks that the embodiment of the present application is provided, is above-mentioned first
Plant and be based on the basis of the data transmission method of RDMA registers memory blocks, also including following technical characteristic：
It is described the bulk memory is divided into multiple memory blocks with default size to be：
The bulk memory is divided into the memory block that multiple sizes are 8K.
Here 8K is a preferred value, however it is not limited to such case, specifically, the size of distribution can be answered according to transmission
Depending on the size of layer data, if each message content of application layer is less than 8K, memory redundancy can be caused, waste memory space,
Should configure smaller；If each message content of application layer is more than 8K, the number of transmissions can be increased and decreased, increase frequency of interaction, it is this
Situation should be configured greatly a bit can be better；That is, the internal memory block size can be defaulted as 8K, this is according to CEPH cluster routines
A median taking of message size.
The 4th kind of data transmission method based on RDMA registers memory blocks that the embodiment of the present application is provided, is above-mentioned first
Plant and be based on the basis of the data transmission method of RDMA registers memory blocks, also including following technical characteristic：
The number of first memory block for being registered to receiving terminal is individual not less than the second memory block for being registered to transmitting terminal
Number is：
The number for being registered to the first memory block of receiving terminal is more than the number of the second memory block for being registered to transmitting terminal.
For example, when transmitting terminal registers 4500 chunk, receiving terminal registers 5000 chunk, this makes it possible to ensure
Receiving terminal has the message that unnecessary chunk sends for receiving end/sending end forever, will not enough make because of the chunk of receiving terminal
Loss of data.
The 5th kind of data transmission method based on RDMA registers memory blocks that the embodiment of the present application is provided, is above-mentioned first
It is also special including following technology kind on the basis of any one in the 4th kind of data transmission method based on RDMA registers memory blocks
Levy：
After the utilization network interface card is by data is activation to be sent to first memory block, also include：
The receiving terminal notifies that the transmitting terminal data transfer is completed, and the memory block that transmitting terminal takes is set into available shape
State.
That is, Infiniband network interface cards provide a kind of ACK mechanism, when the data of transmitting terminal are written to receiving terminal
After in registers memory, receiving terminal is notified that transmitting terminal data transfer is completed, and the chunk that this when, transmitting terminal took is set to can
With state, wait chunk managers to take out and use, thus can further improve the accuracy of data transfer.
The first data transmission device based on RDMA registers memory blocks that the embodiment of the present application is provided is as shown in figure 3, Fig. 3
It is the schematic diagram of the first data transmission device based on RDMA registers memory blocks that the embodiment of the present application is provided, the device bag
Include：
Blocking unit 201, for applying for bulk memory, and is divided into multiple with default size by the bulk memory
Counterfoil is, it is necessary to illustrate, the bulk memory that can apply is dispersed as the memory block (chunk) of 8K mono-；
Registering unit 202, for the memory block to be registered to the transmitting terminal and receiving terminal of network interface card, wherein being registered to reception
The number of first memory block at end is not less than the number of the second memory block for being registered to transmitting terminal, it is necessary to explanation, the network interface card
The network interface card with Infiniband frameworks can be preferably, RDMA is a kind of popular protocol for realizing Infiniband frameworks；
Loading location 203, for obtaining available second memory block, and loads data to be sent；
Transmitting element 204, for utilizing network interface card by data is activation to be sent to first memory block.
Said apparatus can reduce occupancy of the RDMA to memory source, reduce RDMA and taken because calculating operation registers memory
System resource, so as to improve the stability of data transfer, and improve data transmission efficiency.
Second data transmission device based on RDMA registers memory blocks that the embodiment of the present application is provided, is above-mentioned first
Plant and be based on the basis of the data transmission device of RDMA registers memory blocks, also including following technical characteristic：
Specifically for be registered to for memory block using VPI verbs interfaces ibv_reg_mr by the registering unit
The transmitting terminal and receiving terminal of Infiniband network interface cards.
It should be noted that in RDMA communication process, the memory block registered by these is led to the content of transmitting terminal
Cross Infiniband network interface cards and be transferred to receiving terminal, the pressure that data transfer is caused to CPU can be reduced.
The third data transmission device based on RDMA registers memory blocks that the embodiment of the present application is provided, is above-mentioned first
Plant and be based on the basis of the data transmission device of RDMA registers memory blocks, also including following technical characteristic：
The blocking unit by the bulk memory specifically for being divided into the memory block that multiple sizes are 8K.
Here 8K is a preferred value, however it is not limited to such case, specifically, the size of distribution can be answered according to transmission
Depending on the size of layer data, if each message content of application layer is less than 8K, memory redundancy can be caused, waste memory space,
Should configure smaller；If each message content of application layer is more than 8K, the number of transmissions can be increased and decreased, increase frequency of interaction, it is this
Situation should be configured greatly a bit can be better；That is, the internal memory block size can be defaulted as 8K, this is according to CEPH cluster routines
A median taking of message size.
The 4th kind of data transmission device based on RDMA registers memory blocks that the embodiment of the present application is provided, is above-mentioned first
Plant and be based on the basis of the data transmission device of RDMA registers memory blocks, also including following technical characteristic：
The registering unit is registered to reception specifically for the memory block to be registered to the transmitting terminal and receiving terminal of network interface card
The number of first memory block at end is more than the number of the second memory block for being registered to transmitting terminal.
For example, when transmitting terminal registers 4500 chunk, receiving terminal registers 5000 chunk, this makes it possible to ensure
Receiving terminal has the message that unnecessary chunk sends for receiving end/sending end forever, will not enough make because of the chunk of receiving terminal
Loss of data, certainly, this be according to IB network interface cards support chunk register performance indications depending on, while with Installed System Memory size
It is relevant, IB cards point height class, the upper limit that can be born close to equipment and Installed System Memory as far as possible within the scope of.
The 5th kind of data transmission device based on RDMA registers memory blocks that the embodiment of the present application is provided, is above-mentioned first
It is also special including following technology kind on the basis of any one in the 4th kind of data transmission device based on RDMA registers memory blocks
Levy：
Notification unit, for notifying that the transmitting terminal data transfer is completed using the receiving terminal, and transmitting terminal is taken
Memory block be set to upstate.
That is, Infiniband network interface cards provide a kind of ACK mechanism, when the data of transmitting terminal are written to receiving terminal
After in registers memory, receiving terminal is notified that transmitting terminal data transfer is completed, and the chunk that this when, transmitting terminal took is set to can
With state, wait chunk managers to take out and use, thus can further improve the accuracy of data transfer.
In sum, the above method and device management logic are simple, good stability during use, with controllable well
Property；Internal memory fritter disperses, and using flexible, efficiency high, can effectively improve the data transmission efficiency of RDMA communications；RDMA is reduced because of meter
The system resource calculated operation registers memory and take, can be prevented effectively from transmitting terminal and send the problem that internal memory is stepped on, and improve RDMA and lead to
The stability of letter.
The foregoing description of the disclosed embodiments, enables professional and technical personnel in the field to realize or uses the present invention.
Various modifications to these embodiments will be apparent for those skilled in the art, as defined herein
General Principle can be realized in other embodiments without departing from the spirit or scope of the present invention.Therefore, the present invention
The embodiments shown herein is not intended to be limited to, and is to fit to and principles disclosed herein and features of novelty phase one
The scope most wide for causing.
Claims (10)
1. a kind of data transmission method based on RDMA registers memory blocks, it is characterised in that including：
Application bulk memory, and the bulk memory is divided into multiple memory blocks for having and presetting size；
The memory block is registered to the transmitting terminal and receiving terminal of network interface card, wherein being registered to the number of the first memory block of receiving terminal
Not less than the number of the second memory block for being registered to transmitting terminal；
Available second memory block is obtained, and loads data to be sent；
Using network interface card by data is activation to be sent to first memory block.
2. the data transmission method based on RDMA registers memory blocks according to claim 1, it is characterised in that
It is described the memory block is registered to network interface card transmitting terminal and receiving terminal be：
Memory block is registered to the transmitting terminal and receiving terminal of Infiniband network interface cards using VPI verbs interfaces ibv_reg_mr.
3. the data transmission method based on RDMA registers memory blocks according to claim 1, it is characterised in that
It is described the bulk memory is divided into multiple memory blocks with default size to be：
The bulk memory is divided into the memory block that multiple sizes are 8K.
4. the data transmission method based on RDMA registers memory blocks according to claim 1, it is characterised in that
The number of first memory block for being registered to receiving terminal is not less than the number of the second memory block for being registered to transmitting terminal：
The number for being registered to the first memory block of receiving terminal is more than the number of the second memory block for being registered to transmitting terminal.
5. the data transmission method based on RDMA registers memory blocks according to claim any one of 1-4, it is characterised in that
After the utilization network interface card is by data is activation to be sent to first memory block, also include：
The receiving terminal notifies that the transmitting terminal data transfer is completed, and the memory block that transmitting terminal takes is set into upstate.
6. a kind of data transmission device based on RDMA registers memory blocks, it is characterised in that including：
Blocking unit, for applying for bulk memory, and is divided into multiple memory blocks for having and presetting size by the bulk memory；
Registering unit, for the memory block to be registered to the transmitting terminal and receiving terminal of network interface card, wherein being registered to the of receiving terminal
The number of one memory block is not less than the number of the second memory block for being registered to transmitting terminal；
Loading location, for obtaining available second memory block, and loads data to be sent；
Transmitting element, for utilizing network interface card by data is activation to be sent to first memory block.
7. the data transmission device based on RDMA registers memory blocks according to claim 6, it is characterised in that
The registering unit by memory block using VPI verbs interfaces ibv_reg_mr specifically for being registered to Infiniband nets
The transmitting terminal and receiving terminal of card.
8. the data transmission device based on RDMA registers memory blocks according to claim 6, it is characterised in that
The blocking unit by the bulk memory specifically for being divided into the memory block that multiple sizes are 8K.
9. the data transmission device based on RDMA registers memory blocks according to claim 6, it is characterised in that
The registering unit is registered to receiving terminal specifically for the memory block to be registered to the transmitting terminal and receiving terminal of network interface card
The number of the first memory block is more than the number of the second memory block for being registered to transmitting terminal.
10. the data transmission device based on RDMA registers memory blocks according to claim any one of 6-9, its feature exists
In also including：
Notification unit, for notifying that the transmitting terminal data transfer is completed using the receiving terminal, and by transmitting terminal occupancy
Counterfoil is set to upstate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710096131.8A CN106873915A (en) | 2017-02-22 | 2017-02-22 | A kind of data transmission method and device based on RDMA registers memory blocks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710096131.8A CN106873915A (en) | 2017-02-22 | 2017-02-22 | A kind of data transmission method and device based on RDMA registers memory blocks |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106873915A true CN106873915A (en) | 2017-06-20 |
Family
ID=59167661
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710096131.8A Pending CN106873915A (en) | 2017-02-22 | 2017-02-22 | A kind of data transmission method and device based on RDMA registers memory blocks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106873915A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107547623A (en) * | 2017-06-28 | 2018-01-05 | 新华三云计算技术有限公司 | Application data transmission method, device, application server and Application Launcher |
| CN107592361A (en) * | 2017-09-20 | 2018-01-16 | 郑州云海信息技术有限公司 | A kind of data transmission method based on double IB networks, device, equipment |
| CN107908474A (en) * | 2017-10-27 | 2018-04-13 | 郑州云海信息技术有限公司 | A kind of Memory Allocation application method and system based on rdma protocol |
| CN110535811A (en) * | 2018-05-25 | 2019-12-03 | 中兴通讯股份有限公司 | Remote memory management method and system, server-side, client, storage medium |
| CN113923167A (en) * | 2021-09-16 | 2022-01-11 | 深圳致星科技有限公司 | Data transmission optimization method in federal learning |
| CN114153785A (en) * | 2021-11-29 | 2022-03-08 | 北京志凌海纳科技有限公司 | Memory management method and device based on remote direct memory access |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101470665A (en) * | 2007-12-27 | 2009-07-01 | Tcl集团股份有限公司 | Method and system for internal memory management of application system without MMU platform |
| CN101719063A (en) * | 2009-10-30 | 2010-06-02 | 曙光信息产业(北京)有限公司 | Software interface realizing method and device with zero-copy function |
| CN103929415A (en) * | 2014-03-21 | 2014-07-16 | 华为技术有限公司 | Method and device for reading and writing data under RDMA and network system |
| CN103942155A (en) * | 2014-04-29 | 2014-07-23 | 中国科学院微电子研究所 | Memory block control method and device |
-
2017
- 2017-02-22 CN CN201710096131.8A patent/CN106873915A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101470665A (en) * | 2007-12-27 | 2009-07-01 | Tcl集团股份有限公司 | Method and system for internal memory management of application system without MMU platform |
| CN101719063A (en) * | 2009-10-30 | 2010-06-02 | 曙光信息产业(北京)有限公司 | Software interface realizing method and device with zero-copy function |
| CN103929415A (en) * | 2014-03-21 | 2014-07-16 | 华为技术有限公司 | Method and device for reading and writing data under RDMA and network system |
| CN103942155A (en) * | 2014-04-29 | 2014-07-23 | 中国科学院微电子研究所 | Memory block control method and device |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107547623A (en) * | 2017-06-28 | 2018-01-05 | 新华三云计算技术有限公司 | Application data transmission method, device, application server and Application Launcher |
| CN107592361A (en) * | 2017-09-20 | 2018-01-16 | 郑州云海信息技术有限公司 | A kind of data transmission method based on double IB networks, device, equipment |
| CN107592361B (en) * | 2017-09-20 | 2020-05-29 | 郑州云海信息技术有限公司 | Data transmission method, device and equipment based on dual IB network |
| CN107908474A (en) * | 2017-10-27 | 2018-04-13 | 郑州云海信息技术有限公司 | A kind of Memory Allocation application method and system based on rdma protocol |
| CN110535811A (en) * | 2018-05-25 | 2019-12-03 | 中兴通讯股份有限公司 | Remote memory management method and system, server-side, client, storage medium |
| CN110535811B (en) * | 2018-05-25 | 2022-03-04 | 中兴通讯股份有限公司 | Remote memory management method and system, server, client and storage medium |
| CN113923167A (en) * | 2021-09-16 | 2022-01-11 | 深圳致星科技有限公司 | Data transmission optimization method in federal learning |
| CN113923167B (en) * | 2021-09-16 | 2024-05-03 | 深圳致星科技有限公司 | Data transmission optimization method in federal learning |
| CN114153785A (en) * | 2021-11-29 | 2022-03-08 | 北京志凌海纳科技有限公司 | Memory management method and device based on remote direct memory access |
| CN114153785B (en) * | 2021-11-29 | 2022-08-30 | 北京志凌海纳科技有限公司 | Memory management method and device based on remote direct memory access |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106873915A (en) | A kind of data transmission method and device based on RDMA registers memory blocks | |
| CN102473115B (en) | Apparatus and method for efficient data processing | |
| EP2633404B1 (en) | Lock-less buffer management scheme for telecommunication network applications | |
| CN110189121B (en) | Data processing method and device, block chain client and block chain link point | |
| CN102761489B (en) | Inter-core communication method realizing data packet zero-copying based on pipelining mode | |
| CN102473117B (en) | Apparatus and method for memory management and efficient data processing | |
| CN102323894B (en) | System and method for realizing non-blockage mutual calling in distributed application of enterprise | |
| CN101957808B (en) | Communication method among various CPUs (Central Processing Units), system and CPU | |
| CN109032516A (en) | System and method for the enveloping data in expansible storage system agreement | |
| CN101827072A (en) | Virtual memory protocol segmentation offloading | |
| CN103348641A (en) | Method and system for improved multi-cell support on single modem board | |
| CN108055214A (en) | Control the information receiving and transmitting in multi-slot link layer microplate | |
| CN103218266B (en) | The method used when virtual machine communicates with external mechanical, equipment and dummy machine system | |
| US11936571B2 (en) | Reliable transport offloaded to network devices | |
| CN101150487A (en) | A transmission method for zero copy network packet | |
| CN105553887B (en) | The architecture interconnected on tube core | |
| CN101304373A (en) | Method and system for implementing high-efficiency transmission chunk data in LAN | |
| CN109417488A (en) | The method and apparatus of virtual network function resource management | |
| CN108829529A (en) | Virutal machine memory sharing method, device, computer equipment and storage medium | |
| CN107967225A (en) | Data transmission method, device, computer-readable recording medium and terminal device | |
| CN103167037A (en) | Modularized cloud computing system | |
| CN109002363A (en) | A kind of event-handling method, device, equipment and readable storage medium storing program for executing | |
| CN111190854A (en) | Communication data processing method, device, equipment, system and storage medium | |
| US20200348989A1 (en) | Methods and apparatus for multiplexing data flows via a single data structure | |
| CN107154835A (en) | The sending method and device of data |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170620 |