CN110677220A - RDMA (remote direct memory Access) message mechanism based on multi-track redundant response and implementation device thereof - Google Patents
RDMA (remote direct memory Access) message mechanism based on multi-track redundant response and implementation device thereof Download PDFInfo
- Publication number
- CN110677220A CN110677220A CN201910846462.8A CN201910846462A CN110677220A CN 110677220 A CN110677220 A CN 110677220A CN 201910846462 A CN201910846462 A CN 201910846462A CN 110677220 A CN110677220 A CN 110677220A
- Authority
- CN
- China
- Prior art keywords
- message
- receiving
- rdma
- packet
- sending
- 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
- 230000004044 response Effects 0.000 title claims abstract description 119
- 230000007246 mechanism Effects 0.000 title claims abstract description 30
- 230000005540 biological transmission Effects 0.000 claims abstract description 19
- 239000003999 initiator Substances 0.000 claims abstract description 17
- 238000012545 processing Methods 0.000 claims description 10
- 239000000725 suspension Substances 0.000 claims description 8
- 238000012856 packing Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 7
- 238000006467 substitution reaction Methods 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 238000000034 method Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L1/00—Arrangements for detecting or preventing errors in the information received
- H04L1/12—Arrangements for detecting or preventing errors in the information received by using return channel
- H04L1/16—Arrangements for detecting or preventing errors in the information received by using return channel in which the return channel carries supervisory signals, e.g. repetition request signals
- H04L1/1607—Details of the supervisory signal
- H04L1/1635—Cumulative acknowledgement, i.e. the acknowledgement message applying to all previous messages
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L1/00—Arrangements for detecting or preventing errors in the information received
- H04L1/22—Arrangements for detecting or preventing errors in the information received using redundant apparatus to increase reliability
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
- Communication Control (AREA)
Abstract
The invention relates to the technical field of remote main memory access protocols in a high-speed interconnection network, in particular to an RDMA (remote direct memory Access) message mechanism based on multitrack redundancy response and an implementation device thereof. A RDMA message mechanism based on multi-track redundant response comprises 1) a message initiator adopts multi-message parallel split, and data packets of different messages and data packets of the same message are sent out in a disorder way to get on the internet; 2) each request packet of the message is provided with coding information, and a message receiver counts the data volume according to the coding information; 3) the receiving of each request packet does not generate a response packet, when the receiving of the last request packet of the message is completed, the response packet is generated and returned to the message initiator, and the message initiator automatically generates a message completion notice; 4) the response packet of the message is sent through a plurality of network channels in a way of copying a plurality of copies. The RDMA message mechanism provided by the application adopts message-level multi-track redundant response, ensures reliable transmission of messages, reduces the number of response packets, and improves the efficiency of network transmission.
Description
Technical Field
The invention relates to the technical field of remote main memory access protocols in a high-speed interconnection network, in particular to an RDMA (remote direct memory Access) message mechanism based on multitrack redundancy response and an implementation device thereof.
Background
In high performance computing systems, RDMA messaging mechanisms are employed between computing nodes for remote host access, typically integrated in a network interface chip.
Currently, in the field of high-performance computing systems, a commercial network interface chip is basically monopolized by mellonox corporation, and the RDMA mechanism of the network interface chip is implemented by strictly referring to the InfiniBand protocol. The RDMA of the InfiniBand protocol adopts a sliding window mechanism to respond in each packet, each packet is transmitted in order-preserving mode when a plurality of packets of messages are transmitted, the mechanism can generate more response packets, the communication bandwidth is occupied, in addition, the order-preserving transmission requires that the packets can not be out of order when being transmitted on the network, and generally only deterministic routing can be limited and used.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides an RDMA message mechanism based on multi-track redundant response and an implementation device thereof.
The technical scheme adopted by the invention for solving the technical problems is as follows: an RDMA message mechanism based on multi-track redundant response comprises
1) The message initiator adopts multi-message parallel split, and data packets of different messages and data packets of the same message are sent out of order to surf the internet;
2) each request packet of the message is provided with coding information, and a message receiver counts the data volume according to the coding information;
3) the receiving of each request packet does not generate a response packet, when the receiving of the last request packet of the message is completed, the response packet is generated and returned to the message initiator, and the message initiator automatically generates a message completion notice;
4) the response packet of the message is sent through a plurality of network channels in a mode of copying a plurality of copies;
5) and after receiving the response packet, the message initiator performs message matching through the coded information and generates a message completion notification for the matched message.
Preferably, the encoding information includes a message number, a total data amount of the message, a request packet data amount, and target information.
The RDMA message mechanism provided by the application adopts message-level multi-track redundant response, so that the number of response packets is reduced while the reliable transmission of the message is ensured, and the network transmission efficiency is improved; meanwhile, the multi-packet message supports out-of-order transmission, the routing mode is not limited, deterministic routing and self-adaptive routing can be supported, and the network is more flexible to construct.
An apparatus for implementing RDMA message mechanism based on multi-track redundant response comprises
The sending queue module is used for forming the message into an RDMA descriptor and submitting the RDMA descriptor to a sending engine;
the sending engine is used for splitting the message into a plurality of request packets according to the content of the RDMA descriptor, reading the request packets from the memory access interface to obtain data and sending the data to the network interface; and for receiving a response packet from the receive engine;
the receiving engine is used for receiving the request packet from the network interface, generating a response packet and sending the response packet to the network interface, and receiving the response packet from the network interface and sending the response packet to the sending engine; and for writing data to the memory access interface;
a network interface for receiving/transmitting a request packet and receiving/transmitting a response packet;
and the memory access interface is used for receiving a source data read request sent by the sending engine and a target data write request sent by the receiving engine and returning a response.
Preferably, the transmission queue module comprises
A plurality of send queues for writing RDMA descriptors;
and the arbitration unit is used for arbitrating and selecting one RDMA descriptor from the sending queue to submit to the sending engine.
Preferably, the transmission engine comprises
An RDMA descriptor suspension unit to cache RDMA descriptors;
the unpacking and packing unit is used for splitting the message into a plurality of request packets according to the content of the RDMA descriptor and setting unique coding information for each request packet;
the local Get queue is used for sending a request packet data reading request to the memory access management unit;
the memory access management unit is used for sending a source data read request to the memory access interface and receiving a source data read response returned by the memory access interface;
the Put buffer unit is used for reading the request packet from the memory access interface to obtain data and then sending the data to the network interface;
and the response processing unit is used for receiving the response packet from the receiving engine, matching the coding information and generating a message completion notice for the response packet matched with the coding information.
Preferably, the encoding information includes a message number, a total data amount of the message, a request packet data amount, and target information.
Preferably, the receiving engine comprises
A request packet processing unit for receiving the request packet from the network interface and generating a response packet according to the message receiving status; and is used for sending the target data write request to the memory access interface;
a remote ack queue for sending the response packet to the network interface;
the marking unit is used for automatically recording the data volume of the message according to the coding information of the request packet, and triggering to generate the completion of the receiving after the message is received;
and the response packet processing unit is used for receiving the response packet of the network interface and sending the response packet to the sending engine.
Preferably, the network interface comprises
A request/response packet transmitting unit for transmitting the request packet and the response packet to the receiving engine;
and the request/response packet receiving unit is used for receiving the request packet sent by the sending engine and the response packet sent by the receiving engine.
Preferably, the network interface is provided with a plurality of network channels.
Preferably, the memory access interface comprises
And the memory access suspension unit is used for caching a target data write request sent by the receiving engine and a source data read request sent by the sending engine.
The RDMA message mechanism and the implementation device thereof have the advantages that message-level multi-track redundant response is adopted, each message only generates one response packet, the number of the response packets can be obviously reduced, the network transmission efficiency can be improved, and the response packets adopt multi-channel redundant transmission to improve the reliability of message transmission; and the method supports multi-packet messages to be transmitted out of order through multiple channels, has no limitation on the routing mode of the data packet, and can use deterministic routing and self-adaptive routing, thereby enabling the network construction to be more flexible.
Drawings
FIG. 1 is a schematic structural diagram of an RDMA message mechanism implementation apparatus of the present invention;
FIG. 2 is a schematic diagram of a transmit queue module of the apparatus of the present invention;
FIG. 3 is a schematic diagram of the transmit engine of the apparatus of the present invention;
FIG. 4 is a schematic diagram of the structure of the receive engine of the apparatus of the present invention;
FIG. 5 is a schematic diagram of the structure of the network interface of the apparatus of the present invention;
FIG. 6 is a schematic structural diagram of a memory access interface of the apparatus of the present invention.
Detailed Description
The technical scheme of the invention is further explained by the specific implementation mode in combination with the attached drawings.
An RDMA message mechanism based on multi-track redundant response comprises
1) The message initiator adopts the parallel split of multiple messages, and the data packets of different messages and the data packets of the same message are sent out of order to get on the internet.
2) And each request packet of the message is provided with coding information, and the message receiver counts the data volume according to the coding information. The encoding information may include a message number, a total data amount of the message, a request packet data amount, and target information.
3) And when the last request packet of the message is received completely, generating a response packet and returning the response packet to the message initiator, and the message initiator automatically generates a message completion notice.
4) The response packet of the message is sent through a plurality of network channels in a copy and multi-copy mode, and the probability of the response packet loss is reduced in a mode of sending the response packet redundantly.
5) And after receiving the response packet, the message initiator performs message matching through the coded information and generates a message completion notification for the matched message. The response packet carries the message number, the message initiator firstly configures the message number after receiving the message, and the message receiving completion notification is generated if the numbers are matched.
The RDMA message mechanism provided by the application adopts message-level multi-track redundant response, so that the number of response packets is reduced while the reliable transmission of the message is ensured, and the network transmission efficiency is improved; meanwhile, the multi-packet message supports out-of-order transmission, the routing mode is not limited, deterministic routing and self-adaptive routing can be supported, and the network is more flexible to construct.
As shown in FIG. 1, an apparatus for implementing RDMA message mechanism based on multitrack redundancy response comprises
And the sending queue module is used for forming the message into an RDMA descriptor and submitting the RDMA descriptor to the sending engine. The send queue module, as shown in fig. 2, specifically includes multiple send queues for writing RDMA descriptors; and the arbitration unit is used for arbitrating and selecting one RDMA descriptor from the sending queue to submit to the sending engine.
The hardware provides a plurality of sending queues, the software composes the RDMA message requests into RDMA descriptors according to formats and writes the RDMA descriptors into the sending queues, and the arbitration logic arbitrates and selects one RDMA message from the plurality of sending queues to be submitted to a sending engine for further processing.
The sending engine is used for splitting the message into a plurality of request packets according to the content of the RDMA descriptor, reading the request packets from the memory access interface to obtain data and sending the data to the network interface; and for receiving a response packet from the receive engine. The sending engine specifically includes an RDMA descriptor suspension unit, as shown in fig. 3, for caching RDMA descriptors; the unpacking and packing unit is used for splitting the message into a plurality of request packets according to the content of the RDMA descriptor and setting unique coding information for each request packet; the local Get queue is used for sending a request packet data reading request to the memory access management unit; the memory access management unit is used for sending a source data read request to the memory access interface and receiving a source data read response returned by the memory access interface; the Put buffer unit is used for reading the request packet from the memory access interface to obtain data and then sending the data to the network interface; and the response processing unit is used for receiving the response packet from the receiving engine, matching the coding information and generating a message completion notice for the response packet matched with the coding information.
After receiving the RDMA descriptor, the sending engine splits the message into a plurality of network data packets (request packets) according to the content in the descriptor, and the network data packets are sent to the Internet after reading the data from the local main memory. The sending engine can simultaneously process the split of a plurality of messages, and data packets of different messages and data packets of the same message have no sequence requirement and can be simultaneously processed in a concurrent mode. Each internet access data packet (request packet) carries a unique message number, total message data volume, request packet data volume, target information and the like. The sending engine also receives a message response packet (response packet) returned by the target party, and triggers the generation of the message to end for the response packet with the matched message number; and directly discarding the response packet with unmatched message numbers to ensure that the redundant response packet is received only once.
The receiving engine is used for receiving the request packet from the network interface, generating a response packet and sending the response packet to the network interface, and receiving the response packet from the network interface and sending the response packet to the sending engine; and for writing data to the memory access interface. As shown in fig. 4, the receiving engine specifically includes a request packet processing unit, configured to receive a request packet from a network interface, and generate a response packet according to a message receiving status; and is used for sending the target data write request to the memory access interface; a remote ack queue for sending the response packet to the network interface; the marking unit is used for automatically recording the data volume of the message according to the coding information of the request packet, and triggering to generate the completion of the receiving after the message is received; and the response packet processing unit is used for receiving the response packet of the network interface and sending the response packet to the sending engine.
The receiving engine receives the request packet from the network, writes the data into the main memory, records the receiving condition of the message data amount through the identification unit, and generates a message response packet (response packet) to send to the message initiator if the message data receiving is judged to be completed. The identification unit is an independent module and can automatically record the data volume of the message according to the message number, the total data volume of the message and the request packet data volume information carried in the data request packet, and after the message is received, the message is triggered to be received.
And a network interface for receiving/transmitting the request packet and receiving/transmitting the response packet. As shown in fig. 5, the network interface specifically includes a request/response packet sending unit, configured to send a request packet and a response packet to a receiving engine; and the request/response packet receiving unit is used for receiving the request packet sent by the sending engine and the response packet sent by the receiving engine.
Fig. 5 is a network interface structure diagram of two network channel ports, which is not limited to two network channels in practical implementation. The network interface is responsible for uniformly distributing network data packets (request packets) submitted by the sending engine into the two network channels for transmission, and copying two message response packets (response packets) generated by the receiving engine to respectively send and transmit the message response packets through the two network channels. The network interface receives the data packet and the response packet from the two network channels, submits the data packet and the response packet to the receiving engine, and forwards the response packet to the sending engine by the receiving engine.
And the memory access interface is used for receiving a source data read request sent by the sending engine and a target data write request sent by the receiving engine and returning a response. As shown in fig. 6, the memory access interface specifically includes a memory access suspension unit, which is used for caching a target data write request sent by the receiving engine and a source data read request sent by the sending engine.
The memory access interface is divided into a memory access suspension part, an address substitution part, a memory access interface IP part and the like. The main memory write request (target data write request) sent by the receiving engine and the main memory read request (source data read request) sent by the sending engine are cached in the memory access suspension unit, virtual-real substitution is completed through the address substitution unit, then the physical address unit is used for submitting the read-write request to the memory access interface IP, and the memory access interface IP returns response to complete memory access operation.
The RDMA message mechanism realizing device provided by the application adopts message-level multi-track redundant response, each message only generates one response packet, the number of the response packets can be obviously reduced, the network transmission efficiency can be improved, the response packets adopt multi-channel redundant transmission to improve the reliability of message transmission, multi-packet messages are supported to be transmitted out of order through multiple channels, the routing mode of data packets is not limited, deterministic routing and self-adaptive routing can be used, and the network construction is more flexible.
The above-described embodiments are merely illustrative of the preferred embodiments of the present invention and do not limit the spirit and scope of the present invention. Various modifications and improvements of the technical solutions of the present invention may be made by those skilled in the art without departing from the design concept of the present invention, and the technical contents of the present invention are all described in the claims.
Claims (10)
1. An RDMA message mechanism based on multitrack redundancy response (VRRP), characterized in that: comprises that
1) The message initiator adopts multi-message parallel split, and data packets of different messages and data packets of the same message are sent out of order to surf the internet;
2) each request packet of the message is provided with coding information, and a message receiver counts the data volume according to the coding information;
3) the receiving of each request packet does not generate a response packet, when the receiving of the last request packet of the message is completed, the response packet is generated and returned to the message initiator, and the message initiator automatically generates a message completion notice;
4) the response packet of the message is sent through a plurality of network channels in a mode of copying a plurality of copies;
5) and after receiving the response packet, the message initiator performs message matching through the coded information and generates a message completion notification for the matched message.
2. The RDMA message mechanism over multitrack redundancy reply of claim 1, characterized in that: the encoding information includes a message number, a total data amount of the message, a request packet data amount, and target information.
3. An implementation apparatus of RDMA message mechanism based on multitrack redundancy response, characterized in that: comprises that
The sending queue module is used for forming the message into an RDMA descriptor and submitting the RDMA descriptor to a sending engine;
the sending engine is used for splitting the message into a plurality of request packets according to the content of the RDMA descriptor, reading the request packets from the memory access interface to obtain data and sending the data to the network interface; and for receiving a response packet from the receive engine;
the receiving engine is used for receiving the request packet from the network interface, generating a response packet and sending the response packet to the network interface, and receiving the response packet from the network interface and sending the response packet to the sending engine; and for writing data to the memory access interface;
a network interface for receiving/transmitting a request packet and receiving/transmitting a response packet;
and the memory access interface is used for receiving a source data read request sent by the sending engine and a target data write request sent by the receiving engine and returning a response.
4. The apparatus for implementing an RDMA message mechanism based on multitrack redundancy reply according to claim 1, wherein: the sending queue module comprises
A plurality of send queues for writing RDMA descriptors;
and the arbitration unit is used for arbitrating and selecting one RDMA descriptor from the sending queue to submit to the sending engine.
5. The apparatus for implementing an RDMA message mechanism based on multitrack redundancy reply according to claim 1, wherein: the transmission engine comprises
An RDMA descriptor suspension unit to cache RDMA descriptors;
the unpacking and packing unit is used for splitting the message into a plurality of request packets according to the content of the RDMA descriptor and setting unique coding information for each request packet;
the local Get queue is used for sending a request packet data reading request to the memory access management unit;
the memory access management unit is used for sending a source data read request to the memory access interface and receiving a source data read response returned by the memory access interface;
the Put buffer unit is used for reading the request packet from the memory access interface to obtain data and then sending the data to the network interface;
and the response processing unit is used for receiving the response packet from the receiving engine, matching the coding information and generating a message completion notice for the response packet matched with the coding information.
6. The apparatus for implementing an RDMA message mechanism based on multitrack redundancy reply as claimed in claim 5, wherein: the encoding information includes a message number, a total data amount of the message, a request packet data amount, and target information.
7. The apparatus for implementing an RDMA message mechanism based on multitrack redundancy reply according to claim 1, wherein: the receiving engine comprises
A request packet processing unit for receiving the request packet from the network interface and generating a response packet according to the message receiving status; and is used for sending the target data write request to the memory access interface;
a remote ack queue for sending the response packet to the network interface;
the marking unit is used for automatically recording the data volume of the message according to the coding information of the request packet, and triggering to generate the completion of the receiving after the message is received;
and the response packet processing unit is used for receiving the response packet of the network interface and sending the response packet to the sending engine.
8. The apparatus for implementing an RDMA message mechanism based on multitrack redundancy reply according to claim 1, wherein: the network interface comprises
A request/response packet transmitting unit for transmitting the request packet and the response packet to the receiving engine;
and the request/response packet receiving unit is used for receiving the request packet sent by the sending engine and the response packet sent by the receiving engine.
9. The apparatus for implementing an RDMA message mechanism based on multitrack redundancy reply according to claim 8, wherein: the network interface is provided with a plurality of network channels.
10. The apparatus for implementing an RDMA message mechanism based on multitrack redundancy reply according to claim 1, wherein: the memory access interface comprises
And the memory access suspension unit is used for caching a target data write request sent by the receiving engine and a source data read request sent by the sending engine.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910846462.8A CN110677220B (en) | 2019-09-09 | 2019-09-09 | RDMA (remote direct memory Access) message mechanism based on multi-track redundant response and implementation device thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910846462.8A CN110677220B (en) | 2019-09-09 | 2019-09-09 | RDMA (remote direct memory Access) message mechanism based on multi-track redundant response and implementation device thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110677220A true CN110677220A (en) | 2020-01-10 |
| CN110677220B CN110677220B (en) | 2022-06-14 |
Family
ID=69076675
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910846462.8A Active CN110677220B (en) | 2019-09-09 | 2019-09-09 | RDMA (remote direct memory Access) message mechanism based on multi-track redundant response and implementation device thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110677220B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111600872A (en) * | 2020-05-13 | 2020-08-28 | 中国人民解放军国防科技大学 | A memory access legality check controller, chip and device |
| CN112463365A (en) * | 2020-11-13 | 2021-03-09 | 苏州浪潮智能科技有限公司 | Method and device for improving message processing efficiency of Flash channel controller |
| CN114297106A (en) * | 2021-03-02 | 2022-04-08 | 井芯微电子技术(天津)有限公司 | Lumped RDMA link management method |
| CN114827234A (en) * | 2022-04-29 | 2022-07-29 | 广东浪潮智慧计算技术有限公司 | Data transmission method, system, device and storage medium |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8131814B1 (en) * | 2008-07-11 | 2012-03-06 | Hewlett-Packard Development Company, L.P. | Dynamic pinning remote direct memory access |
| US20140164471A1 (en) * | 2006-01-19 | 2014-06-12 | Kenneth G. Keels | Apparatus and method for in-line insertion and removal of markers |
| CN104424105A (en) * | 2013-08-26 | 2015-03-18 | 华为技术有限公司 | Memory data reading and writing processing method and device |
| CN104484295A (en) * | 2014-12-16 | 2015-04-01 | 中国人民解放军国防科学技术大学 | Receiver sliding window-based data transmission method in parallel computer system |
| CN105245470A (en) * | 2015-11-05 | 2016-01-13 | 深圳市网心科技有限公司 | A data transmission method and device |
| US20160080497A1 (en) * | 2011-11-03 | 2016-03-17 | Netapp Inc. | Interconnect delivery process |
| CN105743865A (en) * | 2014-12-12 | 2016-07-06 | 上海大潮电子技术有限公司 | Multimedia file data transmission system and transmission method |
| CN107104782A (en) * | 2017-04-15 | 2017-08-29 | 深圳市民泰科电子有限公司 | Data transmission method and device |
| CN107104902A (en) * | 2017-04-05 | 2017-08-29 | 广东浪潮大数据研究有限公司 | A kind of method, relevant apparatus and the system of RDMA data transfers |
| US20180004705A1 (en) * | 2016-06-29 | 2018-01-04 | Mellanox Technologies Ltd. | Selective acknowledgment of RDMA packets |
| CN109691039A (en) * | 2018-01-16 | 2019-04-26 | 华为技术有限公司 | Message transmission method and device |
| CN109936510A (en) * | 2017-12-15 | 2019-06-25 | 微软技术许可有限责任公司 | Multipath RDMA transfer |
-
2019
- 2019-09-09 CN CN201910846462.8A patent/CN110677220B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140164471A1 (en) * | 2006-01-19 | 2014-06-12 | Kenneth G. Keels | Apparatus and method for in-line insertion and removal of markers |
| US8131814B1 (en) * | 2008-07-11 | 2012-03-06 | Hewlett-Packard Development Company, L.P. | Dynamic pinning remote direct memory access |
| US20160080497A1 (en) * | 2011-11-03 | 2016-03-17 | Netapp Inc. | Interconnect delivery process |
| CN104424105A (en) * | 2013-08-26 | 2015-03-18 | 华为技术有限公司 | Memory data reading and writing processing method and device |
| CN105743865A (en) * | 2014-12-12 | 2016-07-06 | 上海大潮电子技术有限公司 | Multimedia file data transmission system and transmission method |
| CN104484295A (en) * | 2014-12-16 | 2015-04-01 | 中国人民解放军国防科学技术大学 | Receiver sliding window-based data transmission method in parallel computer system |
| CN105245470A (en) * | 2015-11-05 | 2016-01-13 | 深圳市网心科技有限公司 | A data transmission method and device |
| US20180004705A1 (en) * | 2016-06-29 | 2018-01-04 | Mellanox Technologies Ltd. | Selective acknowledgment of RDMA packets |
| CN107104902A (en) * | 2017-04-05 | 2017-08-29 | 广东浪潮大数据研究有限公司 | A kind of method, relevant apparatus and the system of RDMA data transfers |
| CN107104782A (en) * | 2017-04-15 | 2017-08-29 | 深圳市民泰科电子有限公司 | Data transmission method and device |
| CN109936510A (en) * | 2017-12-15 | 2019-06-25 | 微软技术许可有限责任公司 | Multipath RDMA transfer |
| CN109691039A (en) * | 2018-01-16 | 2019-04-26 | 华为技术有限公司 | Message transmission method and device |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111600872A (en) * | 2020-05-13 | 2020-08-28 | 中国人民解放军国防科技大学 | A memory access legality check controller, chip and device |
| CN111600872B (en) * | 2020-05-13 | 2022-05-31 | 中国人民解放军国防科技大学 | Access validity check controller, chip and device |
| CN112463365A (en) * | 2020-11-13 | 2021-03-09 | 苏州浪潮智能科技有限公司 | Method and device for improving message processing efficiency of Flash channel controller |
| CN112463365B (en) * | 2020-11-13 | 2023-01-10 | 苏州浪潮智能科技有限公司 | Method and device for improving message processing efficiency of Flash channel controller |
| US11977935B2 (en) | 2020-11-13 | 2024-05-07 | Inspur Suzhou Intelligent Technology Co., Ltd. | Method and apparatus for improving message processing efficiency of flash channel controller |
| CN114297106A (en) * | 2021-03-02 | 2022-04-08 | 井芯微电子技术(天津)有限公司 | Lumped RDMA link management method |
| CN114827234A (en) * | 2022-04-29 | 2022-07-29 | 广东浪潮智慧计算技术有限公司 | Data transmission method, system, device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110677220B (en) | 2022-06-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11991072B2 (en) | System and method for facilitating efficient event notification management for a network interface controller (NIC) | |
| CN110677220B (en) | RDMA (remote direct memory Access) message mechanism based on multi-track redundant response and implementation device thereof | |
| US6857030B2 (en) | Methods, system and article of manufacture for pre-fetching descriptors | |
| US7571269B2 (en) | Covert channel for conveying supplemental messages in a protocol-defined link for a system of storage devices | |
| KR100675850B1 (en) | NC system with ABI protocol | |
| JP5280135B2 (en) | Data transfer device | |
| US6804673B2 (en) | Access assurance for remote memory access over network | |
| US7996548B2 (en) | Message communication techniques | |
| US7606933B2 (en) | Shared memory and high performance communication using interconnect tunneling | |
| US7308523B1 (en) | Flow-splitting and buffering PCI express switch to reduce head-of-line blocking | |
| US20050132089A1 (en) | Directly connected low latency network and interface | |
| US7460531B2 (en) | Method, system, and program for constructing a packet | |
| EP1581875A2 (en) | Using direct memory access for performing database operations between two or more machines | |
| CN110602211B (en) | Out-of-order RDMA method and device with asynchronous notification | |
| KR20150077288A (en) | A look-aside processor unit with internal and external access for multicore processors | |
| US7421520B2 (en) | High-speed I/O controller having separate control and data paths | |
| CN117015764A (en) | Message communication between integrated computing devices | |
| WO2024227389A1 (en) | Data transmission system, method and apparatus, communication device and storage medium | |
| US8798085B2 (en) | Techniques to process network protocol units | |
| US9201599B2 (en) | System and method for transmitting data in storage controllers | |
| Glass | Fibre Channel: leveraging a commercial networking technology for military applications | |
| WO2022061763A1 (en) | Data storage method and apparatus | |
| US20250028658A1 (en) | System and method for low latency packet processing | |
| US20240223500A1 (en) | Peripheral component interconnect express over fabric networks | |
| CN111200594B (en) | Method, apparatus, data receiving device and storage medium for receiving 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |