KR20200121653A - Method and apparatus for lossless packet transport - Google Patents

Abstract

The present invention relates to an apparatus and method for transmitting a packet transmitted in various application cases performing a mission-critical operation without loss. Provided is a lossless packet forwarding apparatus comprising: a flow lookup block which analyzes a port and packet head of an incoming packet to determine whether to duplicate the packet, and generates a sequence number for each packet of a lossless delivery flow; a header processing block for inserting a first additional header including information according to the determination of the flow lookup block into the packet; a packet replication block for generating information on a packet replication method with reference to the first additional header and inserting a second additional header including the same into the packet; and a packet removal block which discards the sequence number of the received packet if it is the same as the sequence number of the previously received packet.

Description

Method and apparatus for lossless packet transport TECHNICAL FIELD

The present invention relates to a method and apparatus for transmitting a packet without loss in a packet communication network. Specifically, in an LS-based packet network, a mission-critical operation (remote control, remote medical care, etc.), high quality streaming, etc. A lossless packet delivery method and apparatus for transmitting sensitive application packets from a source to a destination without loss.

As the protection switching mechanism currently used to guarantee the reliability of the network in the packet network, OAM & P (Operation, Administration, Maintenance and Protection Switching) technology is used to change the path when a failure occurs by monitoring the state of the delivery path at regular intervals. . According to the conventional technology, if a failure occurs in the working path, the packet transmission path is switched to the backup path within 50ms from the time it is recognized, so packet loss occurs during a certain switching time. Have.

Therefore, when constructing a network infrastructure for smart factories and unmanned vehicles (autonomous vehicles, drones, etc.) that perform mission-deterministic operations based on packet networks, it is essential to provide a lossless delivery function.

In accordance with these requirements, the IEEE 802.1 Working Group, an international standardization organization, completed standardization of Ethernet-based lossless packet delivery technology (FRER, Frame Replication and Elimination for Reliability) for small packet network application, and the IETF DetNet Working Group completed the standardization for medium and large packet network application. IP/MPLS-based lossless packet delivery technology (PREF, Packet Replication and Elimination Function) standardization is being promoted, and related research and development are also actively in progress.

Therefore, early preemption of lossless packet delivery technology is urgently needed in order to gain an advantage over intensifying technology competition.

An object of the present invention is to provide a method and system for transmitting a lossless packet substantially without any loss based on the Ethernet/IP/MPLS packet network technology.

In the lossless packet processing apparatus according to the present invention, a flow lookup block for determining whether to duplicate a packet by analyzing a port and a packet head of an incoming packet, generating a sequence number for each packet of a lossless delivery flow, and determining the flow lookup block A header processing block for inserting a first additional header including information according to the packet into the packet, and by referring to the first additional header to generate information on a packet replication method, and inserting a second additional header including the same into the packet A packet duplication block to perform and a packet removal block for discarding a sequence number of a received packet when the sequence number of a previously received packet is the same.

According to the configuration of the present invention, a method and apparatus for transmitting a lossless packet using packet duplication/deletion can deliver a packet lossless from a source to a destination. Therefore, there is an advantage of being able to safely deliver critical application packets such as mission-critical operation and high-quality streaming through the network.

1 is a conceptual diagram for packet duplication/deletion processing according to the present invention.
2 is a diagram showing an internal format of a packet used in the present invention.
3 is a structural diagram of a lossless packet forwarding apparatus according to the present invention.
4 is a flow chart showing an Ingress packet duplication procedure according to the present invention.
5 is a flow chart showing an egress packet duplication procedure according to the present invention.
6 is a flow chart showing an Ingress packet deletion procedure according to the present invention.
7 is a flowchart illustrating a procedure for deleting an egress packet according to the present invention.
8 is an exemplary system diagram using a lossless packet forwarding apparatus according to the present invention.

Other objects and effects of the present invention, and configurations for achieving them will become apparent with reference to the embodiments of the present specification together with the accompanying drawings.

However, the technical idea of the present invention is not limited to the embodiments disclosed below, but can be implemented in a form in which various changes and modifications have been applied, and the embodiments make the disclosure of the present invention complete, and the technology to which the present invention pertains. It is provided only to completely inform the scope of the invention to those of ordinary skill in the field, and the scope of the present invention is to all changes, modifications, equivalents belonging to the technical idea of the present invention conceived from the embodiments. It should be understood to include substitutes.

Meanwhile, terms used in the present specification are for explaining embodiments and are not intended to limit the present invention. Terms such as “comprise” or “have” are intended to designate the existence of a disclosed feature, number, step, action, component, part, or combination thereof, but one or more other features or numbers, steps, actions, It is to be understood that the possibility of the presence or addition of components, parts, or combinations thereof is not preliminarily excluded.

Likewise, when an element is referred to as being “connected” or “connected” to another element, it may be directly connected or connected to the other element, but other elements may exist in the middle. It should be understood that there is.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs.

Meanwhile, when a certain embodiment can be implemented differently, a function or operation specified in a specific block may occur differently from the order specified in the flowchart. For example, two consecutive blocks may actually be executed at the same time, or the blocks may be executed in reverse depending on a related function or operation.

In describing the present invention, in order to facilitate an overall understanding, the same reference numerals are used for the same elements in the drawings, and duplicate descriptions for the same elements are omitted.

The basic concept of the present invention is based on the Ethernet/IP/MPLS packet network technology, the source end station (or network node) generates and duplicates a packet including sequence number information, and then transmits the duplicated packet through multiple paths, The destination end station (or network node) prevents packet loss by checking the sequence number information of the duplicated packet received through a plurality of paths and deleting the duplicated packet.

1 is a conceptual diagram for explaining the basic concept of the present invention.

A lossless packet forwarding network using packet duplication/deletion consists of end stations (eg terminals) and each network node (eg switches, routers).

The source end station analyzes the incoming packet and, in the case of a service flow that requires lossless transmission, adds sequence number information, which means the sequence of packets, and duplicates it, then adds the route information and replicates to multiple paths. Transmitted packet. Each packet that is copied and transmitted is called a member flow.

The destination end station checks the sequence number information of the incoming packet, processes the first packet among the duplicated packets, and discards the packet received later.

For example, if the source end station duplicates the packet and transmits it over two paths, if the packet of path A is received before the packet of path B due to the difference in propagation delay between the paths, the destination end station is the reception history of the sequence number. Is recorded and managed, and when a packet having the same sequence number is received through path B, the packet is discarded.

Meanwhile, (A) and (B) of FIG. 1 illustrate a case in which a network node performs simple forwarding and a case in which packet duplication/deletion is performed, respectively.

The network node performs a function of performing only simple forwarding of incoming packets and not packet duplication or deletion (Fig. 1(a)), or packet duplication or deletion along with packet forwarding (Fig. 1(b) )). However, in the case of packet duplication in the network node, the same processing as the source terminal station is performed, but sequence number information is not added. That is, the sequence number is generated only at the source terminal station. In the case of packet deletion, the same processing as the destination end station is performed.

2 is a packet format used for lossless packet transmission using packet duplication/deletion.

The packet format uses the IETF MPLS-based DetNet (Deterministic Networking) packet format.

The MPLS-based DetNet packet format consists of a 4-byte S-Label (20-bit Label, 3-bit Experimental, 1-bit Bottom of Stack, 8-bit TTL) to identify service flows that require lossless packet delivery in general MPLS packet format. ), 4-byte DetNet Control Word (d-CW) (consisting of 4-bit 0, 28-bit Sequence Number).

Sequence Number in DetNet Control Word is a field used to duplicate or delete a packet in an MPLS-based packet network. Sequence Number starts from 0 when creating DetNet flow and increases by 1 each time the same flow is transmitted (up to 2 ²⁸ -1).

T-Label is a label for transporting packets according to LSP (Label Switch Path) and is composed of the same bits as S-Label.

Physical refers to a physical layer (eg, Optical Transport Layer (OTN), etc.) header field, and Data Link refers to a data link layer (eg, Ethernet) header field.

Lossless packet forwarding device

3 is a functional block diagram of a lossless packet forwarding apparatus using packet duplication/deletion according to the present invention.

Lossless packet delivery device 1000 using packet duplication/deletion is a flow look-up that processes incoming and outgoing packets with line interface blocks 110, switch interface blocks 135, and processor interface blocks 160 and 170 for external connection. Block (I) 115, header processing block (I) 125, packet removal block (I) 125, packet replication block (I) 130, packet removal block (E) 140, flow lookup It is composed of functional blocks of block (E) 145, header processing block (E) 150, packet replication (E) and output queuing block 155. The device is a hardware programmable Field Programmable Gate Array (FPGA). ) Can be used.

The line interface block 110 performs a connection function with an external network through an Ethernet (eg, nx1GbE or nx10GbE) link.

The switch interface block 135 performs a connection function between a switch fabric card (in the case of a chassis type system) or another packet copy/delete device (in the case of a standalone stand-alone system) through an Interlaken link.

The processor interface blocks 160 and 170 provide a connection function with an external processor for managing look-up tables and registers in the device through an Ethernet (1x1GbE) link or a local bus.

The flow lookup block (I) 115 analyzes the packet port and packet header (Ethernet/IP/MPLS) information introduced through the line interface block 110 to determine the flow ID (Ingress FID, Egress) of the packet for processing the system internal packet. FID), and in the case of a flow requiring lossless delivery, a sequence number is generated, and action flags for packet processing (packet duplication or deletion, MPLS label processing (Push/Pop/Swap)) are determined. Perform. If the MPLS label processing among the action flags is Pop, the decapsulation function of removing the MPLS label is performed.

The header processing block (I) 120 performs a function of inserting an additional header into a packet so that the result of the flow lookup block I is referred to in subsequent blocks to process the packet.

The packet removal block (I) 125 refers to the optional header information of the packet and stores the action flag for the Ingress FID when the packet is deleted, when the sequence number of the packet is different from the previous sequence number, and stores the packet when the packet is the same. Discard. That is, in the case of successive reception of packets having the same sequence number (duplicate reception), the packet is discarded. A packet processing flow chart for packet removal (I) will be described later.

When the action flag for the Ingress FID is packet replication, referring to the optional header information of the packet, the packet replication block (I) 130 provides information on a packet replication method (switch fabric card/line card multicasting). It creates and inserts the packet as an optional header so that it can be processed in a later block.

The packet removal block (E) 140 refers to the optional header information of the packet introduced through the switch interface block, and when the action flag for the Egress FID is packet deletion, the sequence number of the packet is different from the previous sequence number. It checks the sequence history (maximum 4K sequence number reception record) managed in the case, updates the sequence history if there is no reception record, and discards the packet if there is a reception record. A packet processing flow diagram for packet removal (E) will be described later.

The flow lookup block (E) 145 determines the packet output port and encapsulation (Ethernet/MPLS header) information for the Egress FID with reference to the optional header information of the packet.

The header processing block (E) 150 inserts the result of the flow lookup block (E) as an optional header into the packet so that the packet can be processed by referring to the result of the flow lookup block (E) in subsequent blocks, and the packet corresponding to the Egress FID. Performs encapsulation (Etherne/MPLS header) function.

The packet replication (E) and output queuing block 155 refers to the optional header information of the packet, and when the action flag for the EgressFID is packet replication, the output port and encapsulation (Ethernet/MPLS) information of the duplicate packet Is determined, configures an output packet, and performs the function of queuing to the corresponding output queue. If the action flag is not packet duplication, it performs a function of queuing the packet to the corresponding output queue.

Packet replication procedure

Hereinafter, with reference to FIGS. 4 and 5, a packet duplication process of the lossless packet delivery apparatus according to the present invention will be described.

Ingress packet duplication is performed in the packet duplication block (I) 130. As shown in Fig. 4, flow ID (Ingress FID) and packet duplication flag from the optional header of the packet received through the internal data bus first, as shown in FIG. (packetReplication) is extracted (S410).

It is determined whether to perform packet replication by checking whether the packet replication flag is 1 by checking the packet replication flag (S420). The value of the packet duplication flag is set by the header processing block (I) 120 according to a result determined by analyzing port and packet header information of the packet to which the flow lookup block (I) 115 is introduced.

As a result of the determination, if the packet replication flag is 1, the ingress packet replication table lookup is performed using the flow ID as a search key. As a result of the table lookup, the multicasting method and information (switch fabric card/line card) are obtained and configured as an additional header (S430). After that, the packet is transmitted to the switch interface block (S440).

If the packet replication flag (packetReplication) is not 1, it is determined that packet replication is unnecessary and the packet is immediately transferred to the switch interface block (S440).

Fig. 5 is a flow chart of egress packet duplication.

As shown, first, a flow ID (Egress FID) and a packet replication flag (packetReplication) are extracted from the optional header of the packet transmitted through the internal data bus (S510). As described above with reference to FIG. 3, the optional header of the egress packet is determined by the flow lookup block (E) 145 analyzing the port and packet header information of the packet, and the header processing block (E) 150 is It is set.

The packet replication flag (packeReplication) is checked to determine whether to replicate the packet (S520), and if replication is required because the packet replication flag is 1, an egress packet replication table lookup is performed using a flow ID as a search key. As a result of table lookup, the output port and encapsulation (DA, SA, VID, MPLS Labels (T-Label, S-Label)) of the duplicated packet are obtained, and the packet is configured with reference to it (S530), and the corresponding output It is transmitted to the line interface block through the queue (S540).

If the packet replication flag (packetReplication) is not 1, the packet is directly transferred to the line interface block through the output queue corresponding to the output port (S540).

Packet drop procedure

A procedure for removing an ingress packet according to the present invention will be described with reference to FIG. 6.

Ingress packet removal is performed in the packet removal block (I) 125. First, flow ID (Ingress FID), packet deletion flag (packetElimination) from the optional header and d-CW of the packet transmitted through the internal data bus, The member flow ID (ExtID) and the reception sequence number (rxSeqNum) are extracted (S610). The member flow ID is a factor for distinguishing when a member flow for the same packet is introduced to one device.

In order to determine whether to perform packet removal, a packet removal flag is checked (S620).

If the packet removal flag is not 1, the packet is immediately transferred to the next block, that is, the packet replication block (I) 130 (S680).

When the packet removal flag is 1, a table lookup for ingress packet removal processing is performed using Ingress FID and Ext ID as search keys (S630).

As a result of the table lookup, the sequence number (lastRxSeqNum) of the recently received packet is obtained.If the packet is the first received packet among the packets of a newly incoming lossless flow, that is, when a certain lossless flow is completed Since the lossless packet delivery apparatus according to the present invention initializes lastRxSeqNum in the lookup table and sets it to 0xFFF_FFFF, the lastRxSeqNum at the time of receiving the first received packet among newly incoming lossless flows has a value of 0xFFF_FFFF. Therefore, it is determined whether it is the first received packet by checking lastRxSeqNum (S640).

In the case of the first received packet, lastRxSeqNum of the ingress packet removal table is updated to rxSeqNum (S670), and the packet is transferred to the next block, the packet replication block (I) 130 (S680).

If it is not the first received packet, it is checked whether rxSeqNum is the same value as lastRxSeqNum, and if it is the same value, the received packet is discarded because a packet having the same content has already been received (S660).

If rxSeqNum is not the same value as lastRxSeqNum, lastRxSeqNum in the ingress packet removal table is updated to rxSeqNum (S670), and the packet is transferred to the packet replication block (I) (S680). rxSeqNum is the same as the serial number included in the header of the incoming packet.

If it is not the first received packet, it is checked whether rxSeqNum is the same value as lastRxSeqNum, and if it is the same value, the received packet is discarded because a packet having the same content has already been received (S660).

If rxSeqNum is not the same value as lastRxSeqNum, lastRxSeqNum in the ingress packet removal table is updated to rxSeqNum (S670), and the packet is transferred to the packet replication block (I) (S680).

The egress packet removal will be described in detail with reference to FIG. 7.

First, a flow ID (Egres FID) and a reception sequence number (rxSeqNum) are extracted from the optional header and d-CW of the packet transmitted through the internal data bus (S710).

The packet removal flag is checked to determine whether to remove the packet (S715). If the packet removal flag is not 1, the packet is immediately transferred to the flow lookup block E without removing the packet (S760).

When the packet removal flag is 1, a table lookup for egress packet removal processing is performed using the Egress FID as a search key (S720), and the sequence history index (seqHisIdx), which is the result of the egress table lookup, is used as a key for the flow sequence. A table lookup is performed to check the history (up to 4K) (S725).

As a result of the egress table lookup, the sequence number (lastRecovRxSeqNum) of the last approved packet is obtained. In the case of the first received packet, lastRecovSeqNum has a default value of 0xFFF_FFFF. LastRecovRxSeqNum in the egress table plays the same role as lastRxSeqNum in the ingress table and is initialized at the same point. The egress stage collects packets that are ingressed to multiple ports or cards in the device, so the concept of recovery of flows. For clarity, the name of the variable was called lastRecovRxSeqNum.

The lastRecovRxSeqNum value is checked to determine whether it is the first received packet (S730), and if it is the first received packet, the value of the bit 0 position of the sequence history is set to 1 (S730), and the lastRecovSeqNum of the egress packet removal table is set to rxSeqNum. By changing, the sequence history of the sequence history table is updated with the modified history (S755). Then, the packet is transferred to the flow lookup block E (S760). The value of bit 0 of the sequence history indicates whether the most recently recovered sequence number has been received. That is, the sequence history is appropriately shifted according to the sequence number of the packet when a packet is recovered so that bit 0 located at the rightmost position indicates whether a packet with the largest sequence number among the packets received so far is received. It works as much as possible.

If it is not the first received packet, rxSeqNum and lastRecovSeqNum are compared (S740). As a result of the determination, if rxSeqNum is not greater than lastRecovSeqNu, the packet that would have been restored earlier than the most recently restored packet in the flow (the packet sent by the transmitter earlier than the recently restored packet) was only recovered, so Δ (lastReovSeqNum-rxSeqNum) in the sequence history Check the value of the bit position (S745). It is determined whether or not the sequence history value of the Δ bit position is 1 (S750), and if it is 1, since the packet has been previously received, the packet is discarded (S755).

If the sequence history value of the Δ bit position is not 1, the sequence history is updated by changing the Δth bit of the sequence history of the sequence history table to 1 without discarding the packet (S770). Then, the packet is transferred to the flow lookup (E) block (S775).

As a result of the determination of step S740, if rxSeqNum is greater than lastRecovSeqNum, a packet having a sequence number greater than the sequence number of the packet recovered so far in the flow is recovered, so shift the sequence history by Δ(rxSeqNum-lastReovSeqNum) shift left) and change bit 0 to 1 (S760). That is, the Δth bit of the sequence history is set to 1. Thereafter, the lastRecovSeqNum of the egress packet removal table is changed to rxSeqNum (S765), and the sequence history of the sequence history table is updated with the modified history (S770). Then, the packet is transferred to the flow lookup block E (S775).

8 is a diagram showing an exemplary system using a lossless packet forwarding apparatus using packet duplication/deletion according to the present invention.

As an example of a system to which a lossless packet delivery device using packet duplication/deletion is applied, there is a small-capacity stand-alone system or a medium-to-large capacity chassis type system.

The standalone system provides an Ethernet interface (2N x GE/10GE) and consists of two packet copy/delete units and a processor unit. The processor sets or manages the packet duplication/deletion device function. The two packet duplication/deletion devices are connected through a standardized system interface.

The chassis type system consists of a line card, a switch fabric card, and a main processor card. The line card provides an Ethernet interface (N x GE10GE) and consists of a packet copy/delete device, a switch interface fabric device, and a local processor device. Switch fabric cards provide connectivity between multiple line cards (including switch fabric multicasting for packet replication). The main processor card configures or manages devices in the line card and switch fabric card for the packet copy/delete function to operate in the system.

In the above, although the preferred embodiment of the present invention has been described in detail, all the components constituting the illustrated embodiment are described as being combined into one or are described as being combined together to operate, so that the present invention must be such an implementation. It is not limited to examples. That is, within the scope of the object of the present invention, all the constituent elements may be selectively combined and operated in one or more. In addition, although all the components can be implemented as one independent hardware, a program module that performs some or all functions combined in one or more hardware by selectively combining some or all of the components. It may be implemented as a computer program having In addition, such a computer program is stored in a computer readable media such as a USB memory, a CD disk, a flash memory, etc. or an FPGA, and is read and executed by a computer, thereby implementing an embodiment of the present invention. have.

Accordingly, the present invention extends to the configuration defined by the description of the following claims and the scope of their equivalents.

Claims (1)

A flow lookup block that analyzes the port and packet head of the incoming packet to determine whether to duplicate the packet, and generates a sequence number for each packet of a lossless delivery flow;
A header processing block for inserting a first additional header including information according to the determination of the flow lookup block into the packet,
A packet replication block for generating information on a packet replication method with reference to the first additional header and inserting a second additional header including the same into the packet,
A packet removal block that discards when the sequence number of a received packet is the same as that of a previously received packet.
Lossless packet delivery device comprising a.

Images