CN115190132A

CN115190132A - L2TP load scheduling method, device and system

Info

Publication number: CN115190132A
Application number: CN202210770299.3A
Authority: CN
Inventors: 赵红伟; 陈刚; 高婷静
Original assignee: Shanghai Liangxun Electronic Commerce Co ltd
Current assignee: Shanghai Liangxun Electronic Commerce Co ltd
Priority date: 2022-06-30
Filing date: 2022-06-30
Publication date: 2022-10-14
Anticipated expiration: 2042-06-30
Also published as: CN115190132B

Abstract

The invention discloses a method, a device and a system for scheduling L2TP load, wherein the method comprises the following steps: after receiving the message, screening out an L2TP control message data packet, and discarding other packets; extracting LAC and LNS identity information and unique identification marks of connection relation thereof from an L2TP control message data packet; creating a storage table item for the unique identification mark and allocating a storage space; storing the content of each mark successor control message, starting a message retransmission counter and a timer, and recording the retransmission times and the interval duration; analyzing the timing of the timer, calculating the negotiation key data of the Tunnel and the Session when the timer expires, judging the event type according to the calculation result and sending the event type to the instruction module; collecting L2TP configuration of LAC and LNS and sending the configuration to a storage module for storage; the instruction module reads the L2TP configuration and the event type of the LAC and the LNS, and judges whether to send out early warning and whether to send out a configuration modification instruction to the LAC and the LNS. The invention can reduce the Tunnel interruption probability and shorten the recovery time of batch terminal disconnection in the L2TP NAS-Initiated mode.

Description

L2TP load scheduling method, device and system

Technical Field

The invention belongs to the technical field of Internet of things and data communication, and particularly relates to a method, a device and a system for scheduling L2TP (service provider profile) loads.

Background

The layer two tunneling protocol (L2 TP) is a mature VPDN technology, and there are two common tunneling (Tunnel) modes: NAS-Initiated mode and Client-Initiated mode. Wherein, the Tunnel negotiation of the NAS-Initiated mode is Initiated by an L2TP Access Concentrator (LAC: L2TP Access Concentrator), and the terminal only needs to support PPP (Point to Point Protocol) without supporting L2 TP. After the terminal dials in the LAC, the LAC initiates a Tunnel negotiation request to a L2TP Network Server (LNS: L2TP Network Server), session negotiation and ppp dialing are carried out after the Tunnel negotiation is finished, and multi-Session sharing of the Tunnel is supported. The method comprises the steps of establishing, keeping alive and removing Tunnel Control Message (Control Message) queues and periodic keep-alive messages (Hello Message) which depend on UDP format for tunnels (tunnels) of L2TP, wherein each Control Message carries identification attribute information such as Tunnel id and Session id, a pair of Message numbers (Nr: receiving number and Ns: sending number), and LAC and LNS respectively maintain non-conflicted queues for each Tunnel and Session through Control elimination attribute and number values. The LAC and the LNS respectively select whether to send the periodic keep-alive messages, set the sending period, the overtime time and the retransmission times, and consider that the opposite end deletes the tunnel when the Hello mask does not receive the response of the opposite end, so that the tunnel of the local end is also deleted. The problems existing in the prior art are as follows: UDP is a stateless transmission method, and when message transmission is lost, the receiving end and the sending end cannot sense through the UDP protocol, so the LAC and the LNS can only determine to immediately process the message or continue to wait according to whether the respective received message numbers are continuous. In order to avoid too many control messages to be received and transmitted or out-of-sequence numbering, the LAC and the LNS usually set their respective receiving and transmitting windows as buffers for processing by upper modules, and the windows need to occupy storage space. With the development of wireless cellular networks and mobile internet of things, when a large number of terminals of the internet of things access through the NAS-Initiated mode, the number of concurrent tunnels and the number of concurrent sessions in a single Tunnel are much greater than those in a conventional non-internet of things scenario, more dialing and drop events are triggered when the radio environment of the terminal changes and shifts, a large number of control messages triggered by events need to be processed when the Tunnel is maintained by the LAC and the LNS, and the message processing speed and the buffer space are always limited for the LAC and the LNS. In practical application, when a window is full, subsequent messages cannot be received and transmitted, which causes tunnel interruption, resulting in simultaneous disconnection of a large number of terminals, and when a large number of terminals are redialed simultaneously, the LAC and the LNS need to process new sessions as well as clear old sessions, which requires a large amount of hardware computing power and storage space to be consumed in a short time, which may cause the terminal to be redialed repeatedly and cannot be connected, and even may cause system crash of the LAC and the LNS. While the LAC of an ISP will typically serve different clients, APNs, it is also possible for the LNS of the client side to interface with LACs of multiple ISPs. In the prior art, the LAC and the LNS execute load sharing actions based on their own configurations, and cannot be adjusted in real time according to the load state of the opposite end, so that the whole system is affected when the load state of a certain LAC and LNS is severely vibrated. Chinese patent CN102752221B discloses a data packet load sharing method and device applied in L2TP networking, the method comprising: the LAC establishes an L2TP sub-tunnel belonging to the L2TP main tunnel with the LNS according to the routing number of the destination IP address to the L2TP main tunnel, allocates a tunnel source IP address and a tunnel destination IP address to each L2TP sub-tunnel, and sequentially multiplexes the L2TP Session sessions between the LAC and the LNS to the established L2TP sub-tunnels respectively; when the LAC receives a data message in any L2TP Session, the L2TP sub-tunnel multiplexed by the L2TP Session is determined, a public network IP header is packaged for the data message and is forwarded to avoid congestion caused by excessive load of part of links, and the packaged public network IP header comprises a tunnel source IP address and a destination IP address of the L2TP sub-tunnel multiplexed by the L2TP Session. The defects of the invention are that: load sharing scheduling on the LNS side is not involved, and load scheduling based on the Tunnel control message is not involved. In chinese patent CN109257444B, "a load sharing method, device and system", the inventor discloses a load sharing method, device and system, which relates to the field of data communication, and realizes mutual backup between L2TP network servers through a load sharing device, and simultaneously has a load sharing function, thereby reducing the processing pressure of LNS and improving the utilization rate of equipment. The method comprises the following steps: after receiving the session request sent by the at least one LAC, the load sharing device sends the session request to one of the LNS according to a load strategy; after receiving the session response of one of the LNS, the session response is forwarded to the other LNS; and the session response carries session information needing to be backed up. The invention arranges the load sharing device on the communication intermediate link of the LAC and the LNS, thereby becoming a fault point, and simultaneously the invention does not relate to load sharing scheduling on the LAC side and load scheduling based on Tunnel control message. Therefore, how to develop a new L2TP load scheduling method to overcome the above-mentioned drawbacks in the prior art is a direction that needs to be studied by those skilled in the art.

Disclosure of Invention

The invention aims to provide an L2TP load scheduling method, which can reduce the probability of Tunnel interruption and shorten the recovery time of batch terminal disconnection in an L2TP NAS-Initiated mode.

The technical scheme adopted is as follows:

an L2TP load scheduling method comprises the following steps:

step 1: screening out an L2TP control message data packet from the original mirror image flow, and discarding other packets;

and 2, step: extracting the control message type, the LAC and the LNS name from the L2TP control message data packet, and using the control message type, the LAC and the LNS name as a unique identification for identifying the LAC and the LNS identity and relationship;

and step 3: creating a storage table item for the unique identification mark and allocating a storage space;

and 4, step 4: storing the content of each identification successor control message, and recording the retransmission times and duration;

and 5: analyzing the timing of the timer, calculating the negotiation key data of the Tunnel and the Session when the timer expires, and judging the event type according to the calculation result;

step 6: collecting L2TP configuration of LAC and LNS;

and 7: and judging whether to send out early warning and whether to send out a configuration modification instruction to the LAC and the LNS according to the L2TP configuration and the event type of the LAC and the LNS.

Preferably, in the L2TP load scheduling method: in the step 1, the acquisition module screens out the L2TP control message data packet according to the following conditions: the protocol Type is UDP, the destination port is 1701, and Packet Type =1 in the L2TP header.

Preferably, in the L2TP load scheduling method: the unique identification in step 2 includes a source IP, a destination IP, and a Control Message Type value and a Host Name value of the L2TP header in the Control Message packet.

Preferably, in the L2TP load scheduling method: step 3 also includes: and starting an inactivity aging timer while creating the storage table entry, deleting the storage table entry when the inactivity aging timer expires, and recovering the storage space.

Preferably, in the L2TP load scheduling method: and 4, storing the content of each identified subsequent control message, wherein the content comprises a Tunnel ID, a Session ID, ns, nr and a Receive Window Size value in the message.

Preferably, in the L2TP load scheduling method: step 4 also includes: analyzing the load sharing weight of the LAC initiating negotiation to the LNS and the allowed maximum Session number of the LNS based on the times of receiving the negotiation message; and analyzing a retransmission cycle and the maximum retransmission times based on the timestamp of the message when retransmission occurs.

Preferably, in the L2TP load scheduling method: step 5, the negotiation key data of the Tunnel and the Session comprise: active LAC name, active LNS name, active LAC IP number, active LNS IP number, tunnel duration, session number in Tunnel, tunnel establishment success rate, average Tunnel negotiation completion time, session negotiation success rate, average Session negotiation completion time, session cutting frequency, message retransmission rate, message discard rate, window overflow frequency, and message response delay.

In order to implement the above L2TP load scheduling method, the present invention further discloses an L2TP load scheduling device, which includes: the system comprises an acquisition module, a storage module, an analysis module and an instruction module;

the acquisition module is used for screening an L2TP control message data packet from the original mirror image flow, extracting a source IP and a destination IP from the L2TP control message data packet, extracting a control message type, an LAC (local area network) and an LNS (local area network) name from an L2TP header, and sending the control message type, the LAC name and the LNS name to the storage module as identifiers for identifying the unique identity and relationship of the LAC and the LNS; periodically collecting L2TP configuration and other related configuration from the LAC and the LNS and sending the configuration to a storage module;

the storage module is used for storing the received identifiers of the unique identities and the relationships of the LAC and the LNS, storing detailed information of Tunnel and Session in subsequent control messages, and calculating and storing message retransmission times and retransmission time intervals; the method comprises the steps of storing LAC and LNS configuration received from an acquisition module; the system also comprises a calling interface for providing various statistical data of the control message to the analysis module.

The analysis module is used for analyzing the statistical data in the acquired control message and outputting the analysis result to the instruction module;

the instruction module is used for combining and processing the acquired analysis results, judging whether to send a configuration adjustment instruction to LAC and LNS equipment or not according to the acquired LAC and LNS configuration, and editing the sent instruction content.

The invention further discloses an L2TP load scheduling system, which comprises: as described above, the L2TP load scheduler is configured to bypass the communication link between the LAC and the LNS through a forwarding device supporting traffic mirroring.

Compared with the prior art, the method comprises the steps that firstly, the data acquisition module of the dispatcher is used for acquiring the L2TP Tunnel control message in real time; recording negotiation key data of the LAC and the LNS by the storage module; counting the number of lost control messages and the queuing time waiting for being processed through an analysis module of the dispatcher so as to calculate more reasonable tunnel concurrency number and single tunnel session concurrency number; and finally, indicating the LAC and the LNS to dynamically adjust the Tunnel configuration through a scheduler instruction module, and finally realizing the functions of reducing the Tunnel interrupt probability in the L2TP NAS-Initiated mode and shortening the fault recovery time when batch terminals are disconnected.

Drawings

FIG. 1 is a block diagram of a system of embodiment 1;

fig. 2 is a flowchart of the operation of example 1.

The reference numbers correspond to the following part names:

1. an acquisition module; 2. a storage module; 3. an analysis module; 4. and an instruction module.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.

Example 1 as shown in fig. 1-2:

an L2TP load scheduling device and system, the device thereof includes: the system comprises an acquisition module, a storage module, an analysis module and an instruction module;

the acquisition module is used for screening an L2TP control message data packet from the original mirror image flow, extracting a source IP and a destination IP from the L2TP control message data packet, extracting a control message type, an LAC (local area network) and an LNS (local area network) name from an L2TP header, and sending the control message type, the LAC name and the LNS name to the storage module as identifiers for identifying the unique identity and relationship of the LAC and the LNS; periodically acquiring L2TP configuration and other related configuration from the LAC and the LNS and sending the configuration to the storage module;

the instruction module is used for carrying out combination processing on the obtained analysis results, judging whether to send a configuration adjustment instruction to LAC and LNS equipment or not according to the obtained LAC and LNS configuration, and editing the sent instruction content.

The system comprises the following components: as described above, the L2TP load scheduler is configured to bypass the communication link between the LAC and the LNS through a forwarding device supporting traffic mirroring. The forwarding device is a common network device, and the traffic mirroring is a published mature technology, which is not the core of the present application and is not described herein again. The dispatcher establishes an additional communication link with each of the LAC and the LNS through the forwarding equipment to provide channels for configuration acquisition and instruction sending.

The working process is as follows:

an L2TP load scheduling method comprises the following steps:

step 1: screening out an L2TP control message data packet from the original mirror image flow based on the following conditions: "protocol Type is UDP, destination port is 1701, packet Type =1 in L2TP header"; and discarding other packets;

and 2, step: the acquisition module extracts the control message type, the LAC and the LNS name from the L2TP control message data packet, that is: controlling a source IP, a target IP and a Control Message Type value and a Host Name value of an L2TP header in a Message data packet, and sending the source IP, the target IP and the Control Message Type value and the Host Name value as a unique identification for identifying the identity and the relationship of the LAC and the LNS to a storage module;

and step 3: the storage module creates a storage table entry for the unique identification mark, allocates storage space, starts an inactive aging timer, deletes the storage table entry when the inactive aging timer expires, and recovers the storage space; the Tunnel negotiation start and disconnection time stamps are also sent to an analysis module;

and 4, step 4: storing the content of each identified successor control message specifically includes: tunnel ID, session ID, ns, nr, receive Window Size values in the message; analyzing the load sharing weight of the LAC initiating negotiation to the LNS and the maximum allowable number of sessions of the LNS based on the times of receiving the negotiation message; analyzing a retransmission cycle and the maximum retransmission times based on the timestamp of the message when retransmission occurs;

and 5: analyzing the timing of the timer, and calculating the negotiation key data of the Tunnel and the Session when the timer expires, wherein the method specifically comprises the following steps: an active LAC name, an active LNS name, an active LACIP number, an active LNSIP number, a Tunnel duration, a Session number within a Tunnel, a Tunnel establishment success rate, a Tunnel negotiation completion average duration, a Session negotiation success rate, a Session negotiation completion average duration, a Session cut-off number, a message retransmission rate, a message discard rate, a window overflow number, and a message response delay; judging the event type according to the calculation result;

event0: the event is retained. After the last analysis process is completed, the result does not reach the abnormal threshold, which indicates that the analysis module does not detect the abnormality, and is also used for reporting the current working state of the analysis module;

event1: l2TP chain scission. The analysis module receives the expiration of the LAC or LNS inactivity aging timer sent by the storage module and indicates that a certain pair of LAC IP and LNS IP do not send control messages to each other any more;

event2: and (5) cutting off the Tunnel. The storage module receives a Tunnel Control message StopCCN (Stop-Control-Connection-Notification) sent by either the LAC or the LNS, and the analysis module receives a time stamp of the message sent by the storage module, which indicates that one end actively cuts off the Tunnel Connection;

event3: a Tunnel imbalance. The relationship between the number of tunnels established between the LAC and the LNS and the number of IP addresses is unbalanced, which means that the number of tunnels established between the LAC IP address pair and the LNS IP address pair is obviously higher or lower than other IP address pairs;

evnet4: and (4) rejecting the Tunnel. The LNS refuses the LAC to establish a new Tunnel request, which indicates that the number of tunnels allowed by the LNS reaches the upper limit;

event5: session imbalance. Indicating that at least one of the channels established between a certain pair of LAC IP and LNS IP has a Session number obviously higher or lower than that of other channels in one channel;

event6: session rejection. The LNS refuses a request of a LAC for establishing a Session in a certain Tunnel, and the LNS indicates that the number of sessions allowed by the Tunnel reaches the upper limit;

event7: tunnel is unhealthy. The Tunnel negotiation success rate, the Tunnel negotiation completion average time length, the message retransmission rate, the message discarding rate, the window overflow times and the message response time delay, wherein at least one value is higher than a preset threshold, which indicates that the Tunnel state is unstable and the abnormal cut-off possibility is high;

event8: session is unhealthy. The Session negotiation success rate, the average Session negotiation completion duration and the Session cut-off frequency are higher than a preset threshold, which indicates that the Session state in the associated Tunnel is unstable and the terminal frequently drops;

step 6: the acquisition module starts a configuration acquisition timer, acquires L2TP related configuration of the LAC and the LNS when the timer expires, and sends the configuration to the storage module for storage, wherein the acquisition method is not the core of the application and is not described in detail, and the storage module updates after receiving the configuration;

and 7: according to the L2TP configuration and the event type of the LAC and the LNS, whether to send out early warning and whether to send out configuration modification instructions to the LAC and the LNS are judged:

event0: if no alarm exists, no instruction is sent, and the instruction module confirms the working state of the analysis module through the event;

event1: sending an alarm, sending a configuration check instruction to the LAC and the LNS at the same time, checking the routing and negotiation configuration related to the L2 TP;

event2: sending an alarm, sending an instruction of immediately deleting the Session to the LAC and the LNS at the same time, and not waiting for confirmation of an opposite terminal to shorten the cleaning time;

event3: sending an alarm, comparing the configuration related to the number of tunnels established between each IP and LNS acquired from the LAC, and determining whether to instruct the LAC to modify the load sharing configuration related to the IP address pair;

event4: sending an alarm, comparing the configuration related to the Tunnel number established by each IP and LAC obtained from the LNS, determining whether to indicate the LNS to modify the upper limit of the allowed Tunnel number, and determining whether to indicate the LAC to modify the configuration, and no longer sending an invalid Tunnel negotiation request to the IP;

event5: sending an alarm, comparing the number of sessions per Tunnel obtained from the LAC with the load sharing related configuration, and determining whether to instruct the LAC to modify the configuration;

event6: sending an alarm, comparing the configuration related to the Session number of each Tunnel obtained from the LNS, determining whether to indicate the LNS to modify the upper limit of the Session number allowed by the Tunnel, and determining whether to support the LAC to modify the configuration, and no longer sending an invalid Session negotiation request to the Tunnel;

event7; sending an alarm, comparing Tunnel parameter configurations of IP pairs obtained from the LAC and the LNS, sending corresponding configuration modification instructions to the LAC and the LNS, and relaxing Tunnel priority, retransmission, timeout and window size configuration to reduce abnormal cut-off risk;

event8: sending an alarm, comparing the Tunnel parameter configuration of the IP pair obtained from the LAC and the LNS, sending a corresponding configuration modification instruction to the LAC and the LNS, and inhibiting the number of sessions to reduce the message transceiving density;

further preferably, to avoid instruction collision, the instruction module sets a plurality of timers, and a set of completely non-collision instructions is formed by combining a plurality of events received before the timers are overtime and is sent to the LAC and the LNS.

By adopting the technical scheme: according to the method, the L2TP load is scheduled according to the actual working state of the L2TP Nas-Initiated mode, the risk of tunnel interruption is reduced, the fault recovery time of mass terminal disconnection caused by tunnel interruption is shortened, and the stability of L2TP connection under the large connection scene of the cellular Internet of things is improved. The system disclosed by the invention does not depend on the existing LAC and LNS in operation, can be deployed at the LAC side or the LNS side, and can work in a single-point mode or in a distributed mode. The method disclosed by the invention is not in conflict with the existing L2TP load sharing technology, and has the functions of realizing load scheduling based on the actual working state in the prior art and enhancing the rationality of the existing load sharing technology. Because the device does not participate in the forwarding of specific L2TP messages, fault risk points are not added on the basis of a typical L2TP Nas-Initiated mode. The system realizes automatic acquisition, storage, analysis and scheduling, does not need manual intervention, reduces the operation and maintenance workload, and reduces the risk of manual misoperation. According to the performance requirements of the actual application scene on the device, the specification of each module can be adjusted, and the input-output ratio is improved.

Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.

Claims

1. An L2TP load scheduling method is characterized by comprising the following steps:

step 2: extracting the control message type, the LAC and the LNS name from the L2TP control message data packet, and using the control message type, the LAC and the LNS name as a unique identification for identifying the LAC and the LNS identity and relationship;

and step 3: creating a storage table item for the unique identification mark, and allocating a storage space;

and 6: collecting L2TP configuration of LAC and LNS;

and 7: and judging whether to send out early warning and whether to send out a configuration modification instruction to the LAC and the LNS or not according to the L2TP configuration and the event type of the LAC and the LNS.

2. The L2TP load scheduling method of claim 1, wherein the collecting module in step 1 screens out the L2TP control message data packet according to the following conditions: the protocol Type is UDP, the destination port is 1701, and Packet Type =1 in the L2TP header.

3. The L2TP load scheduling method of claim 1, wherein: the unique identification in step 2 includes a source IP and a target IP in the Control Message packet, and a Control Message Type value and a Host Name value of the L2TP header.

4. The L2TP load scheduling method of claim 3, wherein: step 3 further comprises: and starting an inactive aging timer at the same time of creating the storage table entry, deleting the storage table entry when the inactive aging timer expires, and recovering the storage space.

5. The L2TP load scheduling method of claim 1, wherein: and 4, storing the content of each identified subsequent control message, wherein the content comprises a Tunnel ID, a Session ID, ns, nr and a Receive Window Size value in the message.

6. The L2TP load scheduling method of claim 1, wherein: step 4 also includes: analyzing the load sharing weight of the LAC initiating negotiation to the LNS and the allowed maximum Session number of the LNS based on the times of receiving the negotiation message; and analyzing a retransmission cycle and the maximum retransmission times based on the time stamp when the message is retransmitted.

7. The L2TP load scheduling method of claim 1, wherein: step 5, the negotiation key data of the Tunnel and the Session comprise: the method comprises the following steps of an active LAC name, an active LNS name, an active LAC IP number, an active LNS IP number, a Tunnel duration, a Tunnel number of sessions in the Tunnel, a Tunnel establishment success rate, a Tunnel negotiation completion average duration, a Session negotiation success rate, a Session negotiation completion average duration, session cut-off times, a message retransmission rate, a message discard rate, window overflow times and message response time delay.

8. An L2TP load scheduling apparatus, comprising: the system comprises an acquisition module, a storage module, an analysis module and an instruction module;

the storage module is used for storing the received identifiers of the unique identities and the relationships of the LAC and the LNS, storing the detailed information of the Tunnel and the Session in the subsequent control message, and calculating and storing the message retransmission times and the retransmission time interval; the method comprises the steps of storing LAC and LNS configuration received from an acquisition module; the system also comprises a calling interface for providing various statistical data of the control message to the analysis module.

9. An L2TP load scheduling system, comprising: the L2TP load scheduler of claim 8, which is configured to bypass the communication link between the LAC and the LNS via a forwarding device that supports traffic mirroring.