A bypass-mode intelligent media server scheduling method, system, device and medium

Abstract

The present invention belongs to the field of communication technology, and discloses a bypass-mode intelligent server scheduling method, system, device and medium, the method comprising: step one, using the bypass mode to start the detection process, and readjusting the weight proportion of the multi-server architecture according to the resource occupancy rate, hot loading script and early warning model over a period of time; step two, sending data according to the SSDP protocol to adjust the media engine to match different strategy modes to schedule the server. The present invention improves the speed of calculating server strategy selection through the alarm model, improves the server utilization rate, and reduces server crashes; dynamically adjusts the weight value, improves the high availability of the server program group, and maximizes the use of idle servers, reduces service response time, and at the same time obtains IO through the SSDP protocol, calculates CPU energy consumption, and statistically calculates the alarm model data, so as to reasonably use the resources of each server, static stability, and ensure performance utilization efficiency.

Description

Technical Field

[0001] The present invention belongs to the field of communication technology, and in particular relates to a bypass-mode intelligent media server scheduling method, system, equipment and medium. Background Art

[0002] At present, media servers have signaling services, forwarding services, storage services, reasoning services (AI) or other service type servers, collectively referred to as media services, each of which has multiple servers or clusters. In order to reasonably allocate server resources, the original practice is: to find all servers according to the signaling server signaling, and after finding multiple servers, randomly designate a server for forwarding service or storage, reasoning, or directly designate a server for forwarding service according to the weight, or designate a server according to the instant CPU occupancy rate. The weight is generally a normalized number designated by the server, such as server 1: 0.7 server 2: 0.3, then there is a 70% probability of designating server 1 and a 30% probability of designating server 2. The transition probability matrix is ​​[0.7, 0.3].

[0003] Through the above analysis, the problems and defects of the prior art are as follows:

[0004] The actual CPU, GPU, memory, disk, network bandwidth occupancy, disk IO requests, and power consumption of multiple existing servers are constantly changing. Some servers may have occupied more resources for a long time, but because the calculation is too simple (calculated based on the instant occupancy), the weight remains heavy over time, making it impossible to reasonably allocate servers in the presence of media services, AI reasoning services, and storage service architectures. Summary of the invention

[0005] In view of the problems existing in the prior art, the present invention provides a bypass intelligent server scheduling method. The server includes a sip signaling scheduling server and other media servers. When the client connects, it will indicate that it is a publishing identity or a subscription identity, and the signaling scheduling server will schedule and allocate other servers, such as the streaming media server includes a main server and an edge server, the AI ​​reasoning server includes a main reasoning server and an edge reasoning server, and the storage server includes a main storage server and an edge storage server. The weight configured at the beginning of the main server is relatively heavy.

[0006] The present invention is implemented as follows: a bypass intelligent server scheduling method, the bypass intelligent server scheduling method comprising:

[0007] Step 1: Each physical server starts a bypass monitoring process, watch, which detects the CPU, GPU, memory, and disk IO of the physical server. The process sends the detected data according to the SSDP protocol, and the SIP signaling service process obtains the CPU, GPU, IO and other resource occupancy rates according to the SSDP protocol. The server is deleted through the hot loading policy script and early warning model, and the weight ratio of each server is recalculated in the case of multiple servers.

[0008] Hot loading script application process: After the signaling server is started, it will change the event detection script according to the operating system file. When the script file changes, it will re-read the script to obtain the weight formula, thereby updating the server scheduling strategy.

[0009] Step 2: The bypass process watch sends monitoring data according to the SSDP protocol. Each server process adds a script file and executes the function. The amount of code is very small. The script file will send its own function, IP address, etc. The SIP center signaling scheduling server detects how many servers are nearby and the function of each server through the SSDP protocol. At the same time, it changes the scheduling strategy by hot loading the script. The media engine adjusts and matches different strategy modes to schedule the server, thereby scheduling resources in a bypass manner.

[0010] Further, the published data is transmitted to the streaming media server in the form of data stream at the streaming end. After the SIP central signaling scheduling server detects that the data stream has arrived at the streaming media server, it detects the server and its function in the current local area network through the SSDP protocol in the streaming media server. At the same time, the scheduling strategy of the SIP central signaling scheduling server is changed through the Python hot loading script or C++ hot loading script or Lua hot loading script or JS hot loading script prepared in the directory service contained in the inference server. The streaming media server adjusts the corresponding strategy of the matching mode between the origin main service and the edge server according to the scheduling strategy of the SIP central signaling scheduling server. Then the SIP central signaling scheduling server calls the main inference server and the edge inference server in the inference server to perform data stream inference. The storage server stores the processed data stream. The SIP central signaling scheduling server directly calls the data stream in the streaming media server or the data file in the storage server and transmits it to the player in the display play subscription to display the data to the subscriber.

[0011] Further, the strategy mode in step 2 includes bandwidth occupancy type and IO bandwidth type;

[0012] Assume that different servers occupy different network card interfaces;

[0013] The bandwidth occupancy type is a strategy in which the network bandwidth occupancy is high in the weight ratio;

[0014] The bandwidth occupancy rate weight ratio calculation process is as follows:

[0015] X=a*0.4+b*0.3+c*0.1+d*0.1+e*0.1

[0016] In the formula, a is the network bandwidth occupancy, b is the CPU occupancy, c is the GPU occupancy, d is the disk IO occupancy, and e is the memory storage block occupancy;

[0017] The weight ratio calculation process of the IO bandwidth type is as follows:

[0018] X=f / total*0.5+a*0.5

[0019] In the formula, f is the number of disk IO requests, and total is the maximum disk IO request, which can be obtained from the server configuration.

[0020] Furthermore, the specific process of the scheduling is as follows:

[0021] The server calculates the Y value every n (n<=20) seconds, and finds the average value Y / n=X / (60 / n) every minute. Every 5 minutes is a time unit group, which is (Y1, Y2, Y3, Y4, Y5), so only 5 values ​​of 5 minutes can be stored;

[0022] Based on the Y / n values ​​sent by each server, the server with the minimum value is determined as the server with the highest weight.

[0023] Server clusters do not communicate directly with each other, but send data using the SSDP protocol. Each basic weight is dynamically adjusted in the script according to its function.

[0024] Furthermore, the early warning model includes the following data:

[0025] H: cpu warning;

[0026] I: Power usage warning;

[0027] J: Temperature warning;

[0028] K: Insufficient memory warning;

[0029] L: Disk shortage warning;

[0030] Furthermore, the server stores the alarm model data within 5 minutes, and directly deletes the server to be selected according to the early warning model.

[0031] The published data is transmitted to the streaming media server in the form of data streams at the push end. After the SIP central signaling scheduling server detects that the data stream has arrived at the streaming media server, it detects the server and its function in the current LAN through the SSDP protocol in the streaming media server. At the same time, the scheduling strategy of the SIP central signaling scheduling server is changed through the Python hot loading script, C++ hot loading script, Lua hot loading script or JS hot loading script prepared in the directory service contained in the inference server. The streaming media server adjusts the corresponding strategy of the matching mode between the origin main service and the edge server according to the scheduling strategy of the SIP central signaling scheduling server. Then the SIP central signaling scheduling server calls the main inference server and the edge inference server in the inference server to perform data stream inference. The storage server stores the processed data stream. The SIP central signaling scheduling server directly calls the data stream in the streaming media server or the data file in the storage server to transmit it to the player in the display play subscription, and combines the large-screen DLAN protocol to display the data to the subscriber.

[0032] Another object of the present invention is to provide a bypass intelligent server scheduling system for implementing the bypass intelligent server scheduling method.

[0033] The degree system includes:

[0034] A) Customized search discovery protocol ssiop, multicast module, user network protocol, including sending and receiving;

[0035] 1Sending module

[0036] Send the identity to the multicast address 239.255.255.123, port 18000, including its own IP address, name, type, location, CPU occupancy, GPU occupancy, IO occupancy, etc. It is used to indicate its own identity and network address, such as streaming media forwarding service, sip signaling module, AI reasoning module, etc.

[0037] 2 Receiver Module

[0038] Join the multicast address 239.255.255.123, port 18000, receive information, and search for services of interest from various identities, such as finding all streaming media forwarding servers.

[0039] B) Signaling gateway service, used for signaling;

[0040] Supports Sip signaling exchange protocol server, used to record all devices, such as recording all streaming media forwarding servers, and recording all server cpu, gpu, io occupancy and other information. Responsible for notifying stream redirection.

[0041] C) Event warning module, used for event editing and event scripting.

[0042] Write the formula calculation into the event, which can be changed at any time, and even provide an interface for change. Get [ssiop] from the multicast address 239.255.255.123, port 18000, get the occupancy rate, event warning, if it is greater than the warning value, delete the sip service.

[0043] The system provided by the embodiment of the present invention first uses the search protocol SSIOP to search for devices in the current local area network, including camera devices, routing devices, PC devices, and mobile devices, and remotely manages all network devices that support this protocol through the protocol, including monitoring network status, modifying network device configuration, and receiving network event warnings. At the same time, it combines the streaming media transmission protocols RTSP, RTMP, RTP, and RTCP to collect TS and PS streams, transmit the streaming media protocol to the signaling and gateway module, and interact with the AB protocol. Then, the search protocol SSIOP module transmits the collected streaming data to the media engine module. After the SIP center signaling scheduling server detects that the data stream has arrived at the media engine, it detects the server and its functions in the current local area network through the SSIOP protocol, and at the same time, the Python hot loading script or C++ hot loading script or Lua hot loading script or prepared in the directory service contained in the inference server is used. The js hot loading script changes the scheduling strategy of the sip central signaling scheduling server. The media engine adjusts the corresponding strategy of the matching mode between the origin main service and the edge server according to the scheduling strategy of the sip central signaling scheduling server. Then the sip central signaling scheduling server calls the main inference server and the edge inference server in the inference server to perform data flow inference according to the occupancy of GPU, CPU, MEMORY and DISK. The storage server stores the processed data stream. The sip central signaling scheduling server directly calls the data stream in the streaming media server or the data file in the storage server, and transmits it to the html5player or mobileplayer in the display module. Combined with the screen projection protocol, the data is displayed to subscribers. At the same time, if an early warning occurs, the event warning module transmits the event editing data and event script data to the display module in the form of a data stream.

[0044] Another object of the present invention is to provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the bypass intelligent server scheduling method.

[0045] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to perform the steps of the bypass intelligent server scheduling method.

[0046] Another object of the present invention is to provide an information data processing terminal, which is used to implement the intelligent server scheduling system in the bypass mode.

[0047] In combination with the above technical solutions and the technical problems solved, please analyze the advantages and positive effects of the technical solutions to be protected by the present invention from the following aspects:

[0048] The present invention calculates CPU occupancy, memory occupancy, disk block occupancy, IO occupancy and bypass calculation, and uses a ssiop custom protocol to send data, so that servers do not interfere with each other.

[0049] The alarm model of the present invention improves the speed of calculating server strategy selection, improves server utilization, and reduces server crashes; by dynamically adjusting weights, the high availability of server program groups is improved, the use of idle servers is maximized, and service response time is reduced; the ssiop protocol is used to obtain IO, CPU energy consumption calculation, and statistical warnings; at the same time, each server resource is reasonably used, static stability, and performance utilization is guaranteed.

[0050] The expected benefits and commercial value of the technical solution of the present invention after transformation are: automatic discovery, automatic deployment of servers, reduced deployment time, acquisition of logs and various information according to the SSIOP protocol, reduced error detection time, maintenance without multiple command joint checks, all information can be obtained from the SIP server, saving a lot of operation and maintenance time.

[0051] The technical solution of the present invention fills the technical gap in the industry at home and abroad: after the server is started, the server process can be started at any time to join the automatic model, automatically linked to form a server group, and computing services can be provided according to the hot integration script, so that media services, storage services, and reasoning services can be provided automatically without specifying a paired server, such as storage services for media service video streams and AI reasoning services for video streams.

[0052] Does the technical solution of the present invention solve the technical problems that people have been eager to solve but have never been able to solve successfully: greatly reduce the difficulty of operation and maintenance. Intelligent automatic matching service. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a flow chart of the intelligent server scheduling method in bypass mode provided by an embodiment of the present invention;

[0054] Figure 2This is the flow of the data display process provided by the embodiment of the present invention;

[0055] Figure 3 It is a schematic diagram of the signal data processing process of the intelligent server scheduling system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0057] In order to enable those skilled in the art to fully understand how to implement the present invention in detail, this section is an explanatory embodiment that expands and describes the technical solution of the claims.

[0058] The bypass intelligent server scheduling method provided by the embodiment of the present invention includes:

[0059] S101, using a bypass method to start the detection process, and readjust the weight ratio of the multi-server architecture according to the resource occupancy rate, hot loading script and early warning model over a period of time;

[0060] S102, sending data according to the ssiop protocol to enable the media engine to adjust and match different policy modes to schedule the server, so that the server can maximize its efficiency according to its function, and other servers can obtain corresponding data according to the ssdp protocol.

[0061] In the embodiment of the present invention, the policy mode in S102 includes a bandwidth occupancy type and an IO bandwidth type;

[0062] Assume that different servers occupy different network card interfaces;

[0063] The bandwidth occupancy type is a strategy in which the network bandwidth occupancy is high in the weight ratio;

[0064] The calculation process of the weight ratio is:

[0065] X=a*0.4+b*0.3+c*0.1+d*0.1+e*0.1

[0066] In the formula, a is the network bandwidth occupancy, b is the CPU occupancy, c is the GPU occupancy, d is the disk storage block occupancy, and e is the memory storage block occupancy;

[0067] The weight ratio calculation process of the IO bandwidth type is as follows:

[0068] X=f / total*0.5+a*0.5

[0069] In the formula, f is the number of disk IO requests, and total is the maximum disk IO request.

[0070] In the embodiment of the present invention, the specific process of scheduling is:

[0071] The server calculates the Y value every n (n<=20) seconds, and finds the average value Y=X / (60 / n) every minute. Each half hour is a time unit group, which is (Y1, Y2, Y3...Y30), so only 30 half-hour values ​​can be stored;

[0072] Based on the Y / n values ​​sent by each server, the server with the minimum value is determined as the server with the highest weight.

[0073] The server groups do not communicate directly with each other, but send data according to the SSDP protocol. Each basic weight is dynamically adjusted in the script according to different functions. The script has built-in warning signaling. If the specified threshold is exceeded, the server will not be announced offline, but will only announce that it will not accept more service instructions and will continue to serve the connected customer points. If the energy consumption of CPU and other factors remains high (it may be occupied by other processes), the server script will announce that the server is offline, and SIP will delete the server candidate. At the same time, the SIP signaling server returns the streaming server address based on the calculated X value. The streaming server uses the script to calculate the storage server address and the inference server address according to the calculation formula, and notifies the storage server and the inference server.

[0074] If it is a subscription, the subscriber obtains the server address from the SIP server to obtain the stream or store the video file.

[0075] And determine the server weight ratio based on the actual resource occupancy rate. Use hot loading scripts, script change strategy mode, dynamically calculate script formulas, quickly delete server candidates based on early warning information (ssdp protocol), and subdivide server weight ratios based on functions.

[0076] In an embodiment of the present invention, the early warning model includes the following data:

[0077] H: cpu warning;

[0078] I: Power usage warning;

[0079] J: Temperature warning;

[0080] K: Insufficient memory warning;

[0081] L: Disk shortage warning;

[0082] The server stores the alarm model data within 5 minutes and directly deletes the candidate server according to the alarm model.

[0083] In an embodiment of the present invention, the method for the scheduling method to display data to subscribers is:

[0084] The published data is transmitted to the streaming media server in the form of data streams at the push end. After the SIP central signaling scheduling server detects that the data stream has arrived at the streaming media server, it detects the server and its function in the current LAN through the SSDP protocol in the streaming media server. At the same time, the scheduling strategy of the SIP central signaling scheduling server is changed through the Python hot loading script, C++ hot loading script, Lua hot loading script or JS hot loading script prepared in the directory service contained in the inference server. The streaming media server adjusts the corresponding strategy of the matching mode between the origin main service and the edge server according to the scheduling strategy of the SIP central signaling scheduling server. Then the SIP central signaling scheduling server calls the main inference server and the edge inference server in the inference server to perform data stream inference. The storage server stores the processed data stream. The SIP central signaling scheduling server directly calls the data stream in the streaming media server or the data file in the storage server to transmit it to the player in the display play subscription, and combines the large-screen DLAN protocol to display the data to the subscriber.

[0085] Another object of the embodiment of the present invention is to provide a bypass intelligent server scheduling system for implementing the bypass intelligent server scheduling method, the bypass intelligent server scheduling system comprising:

[0086] The media engine module, including the media computing layer and the ssiop protocol, is used to match the origin main service with the edge server;

[0087] Signaling and gateway module, used to interact with the search protocol ssiop module;

[0088] The inference engine module includes the capability AP layer and the ssiop protocol, and is used to perform directory services and main inference services. The directory services include python scripts, C++ scripts, lua scripts, and js scripts.

[0089] Search protocol ssiop module, used to search and discover devices and various occupancy protocols;

[0090] Storage module, including capability AP layer and SSDP protocol, used to store data information;

[0091] Event alert module, used for event editing and event scripting.

[0092] The system provided by the embodiment of the present invention first uses the search protocol SSDP to search for devices in the current local area network, including camera devices, routing devices, PC devices, and mobile devices, and remotely manages all network devices that support this protocol through the SNMP protocol, including monitoring network status, modifying network device configuration, and receiving network event warnings. At the same time, it combines the streaming media transmission protocols RTSP, RTMP, RTP, and RTCP to collect TS and PS streams, transmit the streaming media protocol to the signaling and gateway module, and interact with the AB protocol. Then, the search protocol SSDP module transmits the collected streaming data to the media engine module. After the SIP center signaling scheduling server detects that the data stream has arrived at the media engine, it detects the server and its function in the current local area network through the SSDP protocol, and at the same time, the Python hot loading script or C++ hot loading script or Lua hot loading script prepared in the directory service contained in the inference server is used. The download script or js hot loading script changes the scheduling strategy of the sip central signaling scheduling server. The media engine adjusts the corresponding strategy of the matching mode between the origin main service and the edge server according to the scheduling strategy of the sip central signaling scheduling server. Then the sip central signaling scheduling server calls the main inference server and the edge inference server in the inference server to perform data flow inference according to the occupancy of GPU, CPU, MEMORY and DISK. The storage server stores the processed data stream. The sip central signaling scheduling server directly calls the data stream in the streaming media server or the data file in the storage server, and transmits it to the html5player or mobileplayer in the display module to display the data to the subscriber. At the same time, if an early warning occurs, the event warning module transmits the event editing data and event script data to the display module in the form of data stream.

[0093] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. It can be understood by a person of ordinary skill in the art that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0094] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A bypass intelligent server scheduling method, characterized in that: The bypass intelligent server scheduling method includes: Step 1: Use bypass mode to start the detection process, and readjust the server weight ratio of the multi-server architecture according to resource occupancy, hot loading script and early warning model; Step 2: Send data according to the SSDP protocol to enable the media engine to adjust and match different strategy modes to schedule the server; Each physical server starts a bypass monitoring process watch, which detects the CPU, GPU, memory, and disk IO of the physical server. The process sends the detected data according to the SSDP protocol, and the SIP signaling service process obtains the CPU, GPU, and IO resource occupancy rate according to the SSDP protocol. The server is deleted through the hot loading strategy script and the Markov early warning model, and the weight ratio of each server is recalculated in the case of multiple servers. The hot loading script application process included in the method is: After the signaling server is started, it will change the event detection script according to the operating system file. When the script file changes, it will re-read the script to obtain the weight formula, thereby updating the server scheduling strategy. The strategy modes include bandwidth occupancy type and IO bandwidth type. The scheduling method displays data to subscribers as follows: The published data is transmitted to the streaming media server in the form of data streams at the push end. After the SIP central signaling scheduling server detects that the data stream has arrived at the streaming media server, it detects the server and its function in the current LAN through the SSDP protocol in the streaming media server. At the same time, the scheduling strategy of the SIP central signaling scheduling server is changed through the Python hot loading script, C++ hot loading script, Lua hot loading script or JS hot loading script prepared in the directory service contained in the inference server. The streaming media server adjusts the corresponding strategy of the matching mode between the origin main service and the edge server according to the scheduling strategy of the SIP central signaling scheduling server. Then, the SIP central signaling scheduling server calls the main inference server and the edge inference server in the inference server to perform data stream inference. The storage server stores the processed data stream. The SIP central signaling scheduling server directly calls the data stream in the streaming media server or the data file in the storage server, and transmits it to the player in the display play subscription, and displays the data to the subscriber in combination with the large screen DLAN protocol. The strategy mode in step 2 includes bandwidth occupancy type and IO bandwidth type; Assume that different servers occupy different network card interfaces; The bandwidth occupancy type is a strategy in which the network bandwidth occupancy is high in the weight ratio; The bandwidth occupancy rate weight ratio calculation process is as follows: X=a*0.4+b*0.3+c*0.1+d*0.1+e*0.1 In the formula, a is the network bandwidth occupancy, b is the CPU occupancy, c is the GPU occupancy, d is the disk storage block occupancy, and e is the memory storage block occupancy; The weight ratio calculation process of the IO bandwidth type is as follows: X=(f / total)*0.5+a*0.5 In the formula, f is the number of disk IO requests, and total is the maximum disk IO request.

2. The bypass intelligent server scheduling method according to claim 1, characterized in that: The specific process of the scheduling server is as follows: The server calculates the Y value every n seconds, n<=20, and calculates the average value Y=X / (60 / n) every minute. Each half hour is a time unit group, which is Y1, Y2, Y3...Y30. Only 30 half-hour values ​​are stored. According to the Y / n value sent by each server, the server with the minimum value is determined as the server with the highest weight. According to the alarm model data stored within 5 minutes, the candidate server is directly deleted. The early warning model includes the following data: CPU warning, power usage warning, temperature warning, insufficient memory warning, and insufficient disk warning.

3. An intelligent server scheduling system implementing the bypass mode of the scheduling method according to any one of claims 1-2, characterized in that: The bypass intelligent server scheduling system includes: The media engine module includes the media computing layer and the SSDP protocol, which is used to match the origin main service with the edge server; Signaling and gateway module, used to interact with the search protocol SSDP module; The inference engine module includes the capability AP layer and the SSDP protocol, and is used to perform directory services and main inference services. The directory services include python scripts, C++ scripts, lua scripts, and js scripts. Search protocol SSDP module, used to search protocol SSDP; Storage module, including capability AP layer and SSDP protocol, used to store data information; Display module, including player and large screen DLNA protocol; Event alert module, used for event editing and event scripting.

4. A scheduling method of an intelligent server scheduling system using the bypass method of claim 3, characterized in that: include: First, use the search protocol SSDP to search for devices in the current LAN, including cameras, routers, PC devices, and mobile devices, and remotely manage all network devices that support this protocol through the SNMP protocol, including monitoring network status, modifying network device configuration, and receiving network event warnings. At the same time, combine the streaming media transmission protocols RTSP, RTMP, RTP, and RTCP to collect TS and PS streams, transmit the streaming media protocol to the signaling and gateway modules, and interact with the AB protocol; Then the search protocol SSDP module transmits the collected stream data to the media engine module. After the SIP center signaling scheduling server detects that the data stream has arrived at the media engine, it detects the server in the current LAN and its function through the SSDP protocol. At the same time, it changes the scheduling strategy of the SIP center signaling scheduling server through the prepared Python hot loading script, C++ hot loading script, Lua hot loading script, or JS hot loading script in the directory service contained in the inference server. The media engine adjusts the corresponding strategy of the matching mode between the origin main service and the edge server according to the scheduling strategy of the SIP center signaling scheduling server. Then, the SIP center signaling scheduling server calls the main inference server and the edge inference server in the inference server to perform data flow inference according to the occupancy of GPU, CPU, MEMORY and DISK. The storage server stores the processed data flow. The SIP center signaling scheduling server directly calls the data flow in the streaming media server or the data file in the storage server, and transmits it to the HTML5 player or mobile player in the display module. Combined with the large-screen DLAN protocol, the data is displayed to subscribers. At the same time, if an early warning occurs, the event warning module transmits the event editing data and event script data to the display module in the form of a data stream.

5. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the bypass intelligent server scheduling method as described in any one of claims 1-2.

6. An information data processing terminal, characterized in that: The information data processing terminal is used to implement the intelligent server scheduling system in bypass mode as described in claim 3.

Images