CN112073435B - Method for reducing transmission delay of transmission channel in TOE - Google Patents
Method for reducing transmission delay of transmission channel in TOE Download PDFInfo
- Publication number
- CN112073435B CN112073435B CN202011039238.7A CN202011039238A CN112073435B CN 112073435 B CN112073435 B CN 112073435B CN 202011039238 A CN202011039238 A CN 202011039238A CN 112073435 B CN112073435 B CN 112073435B
- Authority
- CN
- China
- Prior art keywords
- tcp
- function module
- application layer
- protocol function
- segment
- 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.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L69/00—Network arrangements, protocols or services independent of the application payload and not provided for in the other groups of this subclass
- H04L69/16—Implementation or adaptation of Internet protocol [IP], of transmission control protocol [TCP] or of user datagram protocol [UDP]
- H04L69/163—In-band adaptation of TCP data exchange; In-band control procedures
-
- 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
- H04L47/19—Flow control; Congestion control at layers above the network layer
- H04L47/193—Flow control; Congestion control at layers above the network layer at the transport layer, e.g. TCP related
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L69/00—Network arrangements, protocols or services independent of the application payload and not provided for in the other groups of this subclass
- H04L69/16—Implementation or adaptation of Internet protocol [IP], of transmission control protocol [TCP] or of user datagram protocol [UDP]
- H04L69/161—Implementation details of TCP/IP or UDP/IP stack architecture; Specification of modified or new header fields
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Computer Security & Cryptography (AREA)
- Computer And Data Communications (AREA)
Abstract
The invention relates to a method for reducing the transmission delay of a transmitting channel in a TOE, which is characterized in that in the process of downloading an application layer data field from an application layer to a link layer output interface by the TOE transmitting channel, an IP protocol function module and a TCP protocol function module execute the assembly and the downloading of an IP datagram and a TCP message segment without waiting for the completion of the transmission of the application layer data field and the completion of the verification and the calculation of the TCP message segment, and the position of the verification sum of the TCP message segment in a downloading data stream is assigned with an arbitrary value; the ethernet protocol function module starts transmission of the ethernet frame before the ethernet frame is assembled, and can receive the TCP segment checksum from the TCP protocol function module at the time when the TCP segment checksum is assembled in the process of transmitting while assembling. The invention realizes the parallel execution of each functional module of a link layer, a network layer and a transport layer to the maximum extent, thereby reducing the transmission delay of the TOE sending channel.
Description
Technical Field
The invention relates to a method for reducing transmission delay of a sending channel in a TOE (time of arrival), belonging to the technical field of data transmission.
Background
The TCP/IP protocol is a main communication protocol of the internet system and is used in many application scenarios. Implementing the TCP/IP protocol is a work that many networked operating devices must do. The traditional method for realizing the TCP/IP protocol is to run software by using a CPU, but the running software occupies a large amount of system resources, and the software for realizing the TCP/IP protocol is usually run in parallel with software for processing application data carried by the protocol, so that in many cases, especially in the case that the network speed is higher and the data processing task is heavier and heavier in recent years, resource competition between the two causes system performance limitation.
Meanwhile, the running software of the CPU has the characteristics of serial execution of instructions, time division multiplexing of hardware resources, system mechanism restriction by interruption and the like, so that the real-time performance of the realization of the software-based TCP/IP protocol is poor, and the characteristics are specifically represented by large end-to-end delay amount and low predictability of the delay amount. In contrast, the high real-time performance of the communication system is one of the key requirements of many high-performance application scenarios.
The TOE "offloads" the TCP/IP protocol to hardware, that is, implements a so-called TCP/IP Offload Engine (TOE, that is, a "TCP/IP Offload Engine") to make more system resources available to application layer software, and improve the real-time performance of TCP/IP protocol implementation, which is an important research direction in the industry. The TOE expands the TCP/IP protocol stack to transfer part of the TCP/IP protocol from the CPU to the TOE hardware (usually including an ethernet protocol function module located at the link layer, an IP protocol function module located at the network layer, a TCP protocol function module located at the transport layer, and function modules such as the ARP protocol and the ICMP protocol), so as to reduce the burden on the CPU.
As shown in fig. 1 (functional modules such as ARP protocol and ICMP protocol are not shown in the conventional implementation of the TOE sending path), the TCP protocol functional module receives a piece of data downloaded by the application algorithm functional module, stores the piece of data in the buffer memory area, calculates a checksum of a partial field that can be sent in a first TCP segment while receiving an application layer data field, then references the checksum, assembles a TCP segment header according to a TCP segment format, forms a complete TCP segment together with the application layer data field read from the buffer memory area, sends the complete TCP segment to the IP protocol functional module in a byte stream manner, and sends the length of the TCP segment to the IP protocol functional module; the IP protocol functional module calculates the check sum of the IP datagram header by combining the length value of the TCP message segment, assembles the check sum according to the IP datagram message format to form a complete IP datagram header, and sends the complete IP datagram header and the TCP message segment to the Ethernet protocol functional module. The Ethernet protocol functional module assembles the Ethernet frame containing the complete IP datagram according to the Ethernet frame format, calculates a CRC check value, and sends the CRC check value to a data output port of the Ethernet protocol functional module in a byte stream mode.
After the TCP connection is successfully established, there is a transmission delay in the TOE transmission path for the ethernet frame carrying the IP datagram, the TCP segment, and the application layer data. The industry generally defines this amount of transmission delay as: and for a certain section of application layer data field to be sent, the time interval from the time when the last byte of the application layer data field enters the data input interface of the downstream channel of the TOE module to the time when the first byte (excluding the leading field) of the Ethernet frame containing the section of application layer data and the relevant TCP message section and IP datagram leaves the data output interface of the downstream channel of the TOE module. How to reduce the transmission delay amount of the application layer data field in the TOE sending channel becomes a problem to be solved urgently.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: a method is provided that can significantly reduce the amount of transmission delay of an application layer data field in a TOE transmission channel.
In order to solve the technical problems, the technical scheme provided by the invention is as follows: a method for reducing the transmission delay of a sending channel in a TOE (time of arrival), wherein the TOE comprises an Ethernet protocol function module positioned on a link layer, an IP protocol function module positioned on a network layer and a TCP protocol function module positioned on a transport layer, and in the process of sending an Ethernet frame stream, the TCP protocol function module, the IP protocol function module and the Ethernet protocol function module acquire port numbers, IP addresses and MAC addresses of devices on two sides of TCP connection and write TCP message segments, IP datagrams and Ethernet frames to be sent respectively; in the process of downloading an application layer data field from an application layer to a link layer output interface through a TOE sending channel, the IP protocol function module and the TCP protocol function module execute the assembly and the downloading of an IP datagram and a TCP message segment without waiting for the completion of the transmission of the application layer data field and the completion of the calculation of the checksum of the TCP message segment, and the position of the checksum of the TCP message segment in a downloading data stream is assigned with an arbitrary value; the Ethernet protocol function module starts the transmission of the Ethernet frame before the Ethernet frame is assembled, and can receive the TCP segment checksum from the TCP protocol function module before the TCP segment checksum is assembled in the process of assembling and transmitting, thereby ensuring that the conditions of waiting data and further causing message discontinuity in the assembling and transmitting processes caused by starting the transmission of the Ethernet frame before the assembling work of all levels of protocol segments is completed are avoided.
In the prior art, the assembly of segments in each layer of a TOE downlink channel can be started only after the transmission of a data field of a downlink application layer is completed, so that the time when a link layer starts sending an ethernet frame is later than the time when the transmission of the data field of the application layer is completed. The invention realizes the parallel execution of each function module of the link layer, the network layer and the transport layer to the maximum extent by the method of transmitting data (calculating the checksum at the same time) and then transmitting the checksum, so that the link layer starts the sending work of the Ethernet frame when the transmission of the data field of the application layer is not finished, thereby reducing the transmission delay of the TOE sending channel.
It should be noted that: strictly speaking, what is exchanged between the physical layer and the link layer is a bit stream. However, in a specific implementation, a dedicated hardware functional block containing physical layer functions has typically implemented a bidirectional translation function between 8 × N bits and N bytes (N is a natural number), which is exchanged with the link layer by N bytes. Therefore, for the sake of simplicity, the data stream exchanged between layers is referred to as an N-byte stream, i.e., a data stream with a width of N bytes, where N may be any natural number, and the value of N in the "N-byte stream" appearing everywhere in the present invention is not necessarily the same value.
Drawings
The invention will be further explained with reference to the drawings.
Fig. 1 is a conventional implementation of the TOE transmit path.
Detailed Description
Examples
The FPGA has the characteristics of high speed, parallelism, accurate timing and flexible function, and is particularly suitable for realizing TOE. The method of the embodiment can fully exert the characteristics of the FPGA technology, and considers the link layer as a component of the TOE to reduce the transmission delay of the application layer data of the sending channel. Therefore, the present embodiment preferably uses an FPGA to implement the TOE.
In this embodiment, as shown in fig. 1, the TOE includes an ethernet protocol functional module located in a link layer, an IP protocol functional module located in a network layer, and a TCP protocol functional module located in a transport layer (the TOE further includes auxiliary functional modules such as an ARP protocol and an ICMP protocol, but the auxiliary functional modules do not affect the innovation point of the present invention, and are not shown in fig. 1).
In the process of sending the ethernet frame stream by the TOE, after the TCP connection is successfully established, the TCP protocol function module, the IP protocol function module, and the ethernet protocol function module learn the port number, the IP address, and the MAC address of the devices on both sides of the TCP connection, and respectively write the TCP segment, the IP datagram, and the ethernet frame to be sent, which is the prior art, and refer to related documents, and are not described again.
In this embodiment, in the process of downloading an application layer data field from an application layer to a link layer output interface through a TOE sending channel, the IP protocol function module and the TCP protocol function module perform assembly and downloading of an IP datagram and a TCP message segment without waiting for completion of transmission of the application layer data field and completion of computation of a TCP message segment checksum, and assign an arbitrary value to a position where the TCP message segment checksum in a downloading data stream is located; the Ethernet protocol function module starts the transmission of the Ethernet frame before the Ethernet frame is assembled, and can receive the TCP segment checksum from the TCP protocol function module before the TCP segment checksum is assembled in the process of assembling and transmitting, thereby ensuring that the conditions of waiting data and further causing message discontinuity in the assembling and transmitting processes caused by starting the transmission of the Ethernet frame before the assembling work of all levels of protocol segments is completed are avoided.
In this embodiment, a data transmission process of the TOE sending channel is described in detail with reference to fig. 1:
after the TCP connection is successfully established, the TCP protocol function module, the IP protocol function module, and the ethernet protocol function module in the local TOE sending channel can obtain the port number, the IP address, and the MAC address of the devices on both sides of the TCP connection, and then write the port number, the IP address, and the MAC address into the TCP segment structure, the IP datagram structure, and the ethernet frame structure to be sent, respectively, together with other known information.
The application algorithm function module knows that the length of the application layer data field to be transmitted is L0 before the application algorithm function module downloads the application layer data field, and sends the length value L0 to the TCP protocol function module in a time period which is no later than the time when the application layer data field starts to be downloaded.
After receiving the length value of the application layer data field, the TCP protocol function module first needs to confirm that the size of the available space of the buffer area for temporarily storing the received application layer data field, which is obtained by the TCP protocol-specified information exchange mechanism, on the opposite side of the TCP connection is larger than the length value (named as "acknowledgement a" here), and if otherwise, the subsequent sending-related action cannot be started, the received application layer data field can only be stored in the local buffer storage area.
After confirming a, the TCP protocol function module compares L0 with an upper limit Lmax of a length value of an application layer data field carried by a known single TCP segment on the local side to determine a length L1 of the application layer data field that can be carried in a TCP segment carrying the application layer data field (if L0 is less than or equal to Lmax, L1= L0, L0 is greater than Lmax, L1= Lmax), and further determines a length L2 of a TCP segment carrying the application layer data field (L2 = L1+ H1, H1= TCP segment header length) in combination with a current working state of the module, sends a value of L2 to the IP protocol function module, and calculates a checksum of the TCP segment.
After receiving the value of L2, the IP protocol function module determines, in conjunction with the current operating state of the module, the length L3 of the IP datagram (L3 = L2+ H2, H2= IP segment header length) used to carry this TCP segment, and then sends the value of L3 to the ethernet protocol function module and uses it to calculate the header checksum of this IP datagram.
After receiving the value of L3, the ethernet protocol function module, in conjunction with the current operating state of the module, determines length L4 of the ethernet frame (excluding the preamble field and the CRC check field) used to carry the IP datagram (L4 = L3+ H3, H3= data length from the start byte of the ethernet frame excluding the preamble field to the last byte before the IP datagram).
From the ethernet frame format, the IP datagram format, the TCP segment format, the parameter D1 can be determined by each module in real time (D1 = data length from the start byte of the ethernet frame to the last byte before the checksum field in the TCP segment).
Let S = the data rate of the N byte stream transmitted between the layers of the TOE under normal conditions, then: the time interval from the first byte of the ethernet frame to be sent (excluding the preamble field) to the TCP segment "checksum" field to be sent T1= D1/S.
Let T2= the time interval required by the TCP protocol functional module from the time (named T0) when the application layer algorithm functional module finishes receiving the application layer data field with the length L1 to the time when the TCP segment checksum calculation is completed, then, with the time T0 as the reference time, the time when the sending of the first byte of the ethernet frame (excluding the preamble field) is started may be determined as Ts = T0-T1+ T2, that is, ts + T1= T0+ T2, and after the sending is started, the TCP segment checksum needs to be sent after the time period T1 elapses, at this time, it is also the time T0+ T2, that is, the time when the TCP segment checksum calculation is completed, and at this time, the ethernet protocol functional module may directly write the value of the TCP segment checksum into the position in the sending sequence where it should be located.
In fact, as long as the design is reasonable, T1 is greater than T2, that is, the Ts time will be earlier than the T0 time by T1-T2.
Because of the unidirectionality of the time line, it is impossible to determine the time of Ts by T0 reverse estimation in actual operation, so in this embodiment, with the time Td at which the TCP protocol functional module starts to receive the application layer data field as a reference, ts is obtained by forward estimation, and the method is as follows:
let Sd = the data rate of transmitting the application layer data field from the application algorithm functional module to the TCP protocol functional module, and then, since T0-Td = L1/Sd, ts = T0-T1+ T2= Td + L1/Sd-D1/S + T2, wherein the arrival of the time Td may be determined at the time when the TCP protocol functional module starts receiving the application layer data field, L1 may be determined at the latest after Td, and Sd, S, D1, T1, and T2 may be determined according to the message format, the design parameter of TOE, and the current state specified by the TCP/IP protocol.
And the TCP function module stores the application layer data field with the length of L1 in a buffer storage area in the process of receiving the application layer data field with the length of L1, and calculates the checksum of the application layer data field. If it is determined that the size of the available space of the buffer used by the TCP connection peer device to temporarily store the received application layer data field (which has been obtained by the information exchange mechanism specified by the TCP protocol before) is larger than the length of the application layer data field received this time by the TOE transmission channel (i.e. the aforementioned "acknowledgement a" slightly later than Td), the ethernet protocol function module may be caused to start (named "start a") a transmission preparation process of an ethernet frame including the aforementioned application layer data field with the length L1, even if the application layer data field is still in the process of being transmitted to the TCP protocol function module.
From the time start-up a is issued: the TCP protocol function module assembles a TCP segment header according to a TCP segment format while receiving the application layer data field and calculating the checksum, and the TCP segment header and the application layer data field read from the buffer storage area form a complete TCP segment and are sent to the IP protocol function module in an N byte stream mode, wherein the field of the position of the TCP segment checksum is assigned with any value;
the IP protocol function module refers to the value of L3, calculates according to the format of the IP datagram message to obtain the checksum of the IP datagram header, assembles the IP datagram header, and sends the checksum and the TCP message segment to the Ethernet protocol function module;
the Ethernet protocol function module assembles an Ethernet frame containing a complete IP datagram according to an Ethernet frame format, calculates a CRC value of the Ethernet frame, and sends the Ethernet frame to a special hardware interface positioned at an output interface of a link layer in an N byte stream form from the moment Ts;
because the Ts moment is obtained by the derivation process, as long as the related implementation process is not problematic, when the process of assembling and downloading the Ethernet frame by the Ethernet protocol function module is carried out to the position of the TCP segment checksum, the TCP segment checksum can be received from the TCP protocol function module and assembled into the Ethernet frame, thereby ensuring that the condition that the message is discontinuous because the transmission of the Ethernet frame is started before the assembly work of all levels of protocol segments is completed to cause the data waiting in the assembling and transmitting process can not occur.
The present invention is not limited to the specific technical solutions described in the above embodiments, and other embodiments may be made in the present invention in addition to the above embodiments. It will be understood by those skilled in the art that various changes, substitutions of equivalents, and alterations can be made without departing from the spirit and scope of the invention.
Claims (1)
1. A method for reducing transmission delay of a sending channel in a TOE (time of arrival), wherein the TOE comprises an Ethernet protocol function module positioned on a link layer, an IP protocol function module positioned on a network layer and a TCP protocol function module positioned on a transport layer, and in the process of sending Ethernet frame flow of the TOE, the TCP protocol function module, the IP protocol function module and the Ethernet protocol function module acquire port numbers, IP addresses and MAC addresses of devices on two sides of TCP connection and write TCP message segments, IP datagrams and Ethernet frames to be sent respectively; the method is characterized in that:
in the process of downloading an application layer data field from an application layer to a link layer output interface through a TOE sending channel, the IP protocol function module and the TCP protocol function module execute the assembly and the downloading of an IP datagram and a TCP message segment without waiting for the completion of the transmission of the application layer data field and the completion of the calculation of the checksum of the TCP message segment, and the position of the checksum of the TCP message segment in a downloading data stream is assigned with an arbitrary value; the Ethernet protocol functional module starts the sending of the Ethernet frame before the Ethernet frame is assembled, and can receive a TCP segment checksum from the TCP protocol functional module before the TCP segment checksum is assembled to the position of the TCP segment checksum in the process of assembling and sending the Ethernet frame;
let L1 be the length of the application layer data field carried in the TCP segment,
s is the rate at which data is transferred between layers of the TOE,
sd is the rate at which the TCP protocol function receives the application layer data field,
d1 is the data length from the first byte of the ethernet frame that does not include a preamble field to the last byte in the TCP segment that precedes the checksum,
t0 is the moment when the TCP functional module finishes receiving the application layer data field with the length of L1,
t1 is the time interval between the moment when the first byte of the ethernet frame without the preamble field is to be sent and the moment when the checksum field of the TCP segment is to be sent, T1= D1/S;
t2 is the time interval from time T0 until the TCP segment checksum calculation is completed,
td is the time when the TCP protocol function starts receiving the application layer data field,
ts is the moment of starting to send the first byte of the Ethernet frame without the leader field;
then, since T0-Td = L1/Sd, ts = T0-T1+ T2, T1= D1/S, we obtain: ts = Td + L1/Sd-T1+ T2= Td + L1/Sd-D1/S + T2;
the TCP protocol function module stores the application layer data field with the length of L1 in a buffer storage area in the process of receiving the application layer data field, and calculates the checksum of the application layer data field;
before the TCP protocol function module starts to receive the data field of the application layer, according to the format of the TCP message segment and the known TCP link information, the TCP protocol function module completes the assembly work of a part of the header of the TCP message segment and completes a part of the operation work in the detection and calculation of the TCP message segment;
the TCP protocol function module assembles a TCP segment header according to a TCP segment format while receiving the application layer data field and calculating the checksum of the application layer data field, and forms a complete TCP segment together with the application layer data field read from the buffer storage area and sends the complete TCP segment to the IP protocol function module in an N byte stream mode;
the IP protocol function module calculates to obtain an IP datagram header checksum according to the IP datagram message format, assembles the IP datagram header and sends the IP datagram header checksum and the TCP message segment to the Ethernet protocol function module;
from the time of Ts, the Ethernet protocol functional module assembles an Ethernet frame containing the IP datagram according to the Ethernet frame format, calculates the CRC value of the Ethernet frame and sends the Ethernet frame in the form of N byte streams.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011039238.7A CN112073435B (en) | 2020-09-28 | 2020-09-28 | Method for reducing transmission delay of transmission channel in TOE |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011039238.7A CN112073435B (en) | 2020-09-28 | 2020-09-28 | Method for reducing transmission delay of transmission channel in TOE |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112073435A CN112073435A (en) | 2020-12-11 |
| CN112073435B true CN112073435B (en) | 2022-11-04 |
Family
ID=73684372
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011039238.7A Active CN112073435B (en) | 2020-09-28 | 2020-09-28 | Method for reducing transmission delay of transmission channel in TOE |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112073435B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114138707B (en) * | 2021-11-30 | 2023-02-21 | 江南大学 | Data transmission system based on FPGA |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5815516A (en) * | 1996-04-05 | 1998-09-29 | International Business Machines Corporation | Method and apparatus for producing transmission control protocol checksums using internet protocol fragmentation |
| EP1124362A2 (en) * | 2000-01-19 | 2001-08-16 | Wiznot Corp. | Apparatus for processing TCP/IP by hardware, and operating method therefor |
| CN110109852A (en) * | 2019-04-03 | 2019-08-09 | 华东计算技术研究所（中国电子科技集团公司第三十二研究所） | System and method for realizing TCP _ IP protocol by hardware |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7502925B2 (en) * | 2004-04-19 | 2009-03-10 | Nvidia Corporation | Method and apparatus for reducing TCP frame transmit latency |
-
2020
- 2020-09-28 CN CN202011039238.7A patent/CN112073435B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5815516A (en) * | 1996-04-05 | 1998-09-29 | International Business Machines Corporation | Method and apparatus for producing transmission control protocol checksums using internet protocol fragmentation |
| EP1124362A2 (en) * | 2000-01-19 | 2001-08-16 | Wiznot Corp. | Apparatus for processing TCP/IP by hardware, and operating method therefor |
| CN110109852A (en) * | 2019-04-03 | 2019-08-09 | 华东计算技术研究所（中国电子科技集团公司第三十二研究所） | System and method for realizing TCP _ IP protocol by hardware |
Non-Patent Citations (1)
| Title |
|---|
| 用于TCP/IP减荷的智能网卡的设计与实现;杨松岸等;《计算机工程》;20040731;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112073435A (en) | 2020-12-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7912064B2 (en) | System and method for handling out-of-order frames | |
| US10404625B2 (en) | Ethernet enhancements | |
| US7397800B2 (en) | Method and system for data placement of out-of-order (OOO) TCP segments | |
| US8677010B2 (en) | System and method for TCP offload | |
| US7385923B2 (en) | Method, system and article for improved TCP performance during packet reordering | |
| JP7142722B2 (en) | Transmission control method and device | |
| US11012367B2 (en) | Technologies for managing TCP/IP packet delivery | |
| JP2000224261A (en) | Data link control protocol directly supporting network layer protocol and its method | |
| US9130957B2 (en) | Data communication apparatus and method | |
| CN112073435B (en) | Method for reducing transmission delay of transmission channel in TOE | |
| CN114631297A (en) | Method and network device for multipath communication | |
| EP1460804B1 (en) | System and method for handling out-of-order frames (fka reception of out-of-order tcp data with zero copy service) | |
| US11245635B2 (en) | Feedback loop for frame maximization | |
| CN112073436B (en) | Method for reducing transmission delay of receiving channel in TOE | |
| CN115714826A (en) | AF _ XDP-based GigE Vision protocol data packet processing method and system | |
| US7502324B1 (en) | TCP retransmission and exception processing in high speed, low memory hardware devices | |
| JP6933207B2 (en) | Transmitter, method and program | |
| JP2009033676A (en) | Communicating system, communication device and packet transmission method | |
| TWI484793B (en) | Communication device and packet processing method | |
| CN112073434B (en) | Method for reducing TOE-based transmission delay of receiving channel of high-frequency transaction terminal | |
| JP4510524B2 (en) | Data communication device | |
| CN116545948A (en) | Reliable communication method based on UDP protocol | |
| CN113630338A (en) | Method, system, equipment and medium for processing SMB request | |
| JP2013046173A (en) | Control apparatus, control method and program | |
| CN117221417A (en) | TCP/IP protocol unloading engine device |
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 |