Method for improving single-machine throughput in estimated peak period

Abstract

The invention discloses a method for improving throughput of a single machine in an estimated peak period, which comprises the following steps of: when the single machine is close to the bottleneck of a system, performing degradation configuration through a big data algorithm model according to vehicle types, filtering out non-common vehicle types, and only estimating common vehicle types, so that the system can provide normal service under the condition that the capacity is not expanded in the peak period; the platform can bear higher flow; s1, the pre-estimation service configures a degraded vehicle type according to a common vehicle type provided by big data, and configures a system initialization threshold according to a pressure measurement result; s2, the upstream makes a multi-vehicle-type estimation request for the estimation service; s3, the system checks whether the estimated service is in a degradation state or not; through technical means, configuration of available vehicle types is adjusted and estimated, only common vehicle types are estimated, and non-common vehicle types are filtered out, so that system resources are saved, and single-machine throughput is improved; the platform income is increased, the use experience of drivers and passengers can be improved, and the public praise of companies is indirectly improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of online car-hailing, and specifically relates to a method for improving the estimated peak-period single-machine throughput. Background Art

[0002] During holidays, rush hour, or rainy and snowy weather, the number of passenger orders will surge, and it is found that most estimates are for five or six models at the same time, but passengers only choose two or three of them when placing an order. If the estimation service fails to provide effective warning and timely capacity expansion when traffic suddenly increases, the service will not be able to handle the surge in traffic, resulting in an inability to estimate, and passengers will be unable to place orders, resulting in a decline in the experience of passengers and drivers, and the platform will also lose revenue, affecting the company's reputation.

[0003] To this end, the present invention designs a method for improving the estimated peak-period single-machine throughput. Summary of the invention

[0004] The object of the present invention is to provide a method for improving the estimated peak-period single-machine throughput to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above purpose, the present invention provides the following technical solutions: a method for improving the estimated peak single-machine throughput. When it is found that a single machine is close to the system bottleneck, the big data algorithm model can be used to downgrade the configuration according to the vehicle model, filter out uncommon models, and only estimate the commonly used models, so that the system can provide normal services without capacity expansion during peak periods, so that the platform can undertake higher traffic; the specific steps are as follows:

[0006] S1: The estimation service configures downgraded models based on the common models provided by big data, and configures the system initialization threshold based on the stress test results;

[0007] S2: The upstream makes a multi-car estimation request to the estimation service;

[0008] S3: The system checks whether the estimated service itself is in a degraded state;

[0009] S4: The estimation service obtains resource monitoring information in real time before processing the estimation, including system thread pool usage and CPU usage, and updates the system threshold in real time;

[0010] S5: The estimation service determines whether the system throughput reaches the threshold;

[0011] S6: If the estimated service throughput reaches the threshold, a synchronization lock is added to write the service degradation status. The service status is modified with the keyword volatile to ensure thread visibility. At the same time, a delayed task is started to update the service status to normal.

[0012] S7: If the estimated service is in a downgraded state, check whether the vehicle model downgrade filter is configured and whether the estimated vehicle model includes a downgraded vehicle model;

[0013] S8: If vehicle type downgrade filtering is configured and the estimated vehicle type includes downgraded models, the estimation parameters are reorganized to estimate only the downgraded models;

[0014] S9: The estimation service performs normal estimation and returns;

[0015] S10: It is estimated that the service delay task will recover from the degraded state after a certain period of time.

[0016] Preferably, the vehicle type downgrade configuration file used is configured as follows:

[0017] {

[0018] "groupLimit":{

[0019] "useLimit":,

[0020] "sizeLimit":,

[0021] "sizeLimitTimeout":,

[0022] "goupLimitSet":[],

[0023] "channelGroupLimitMap":{

[0024] "partner-1":[],

[0025] "partner-2":[]

[0026] }

[0027] }

[0028] }

[0029] The above configuration description:

[0030] useLimit: whether to use the estimated vehicle model downgrade filtering, including true to use, false not to use;

[0031] sizeLimit: system initialization threshold. Downgrade occurs when this threshold is reached. The system automatically updates this threshold based on internal resource monitoring.

[0032] sizeLimitTimeout: system downgrade time window. All requests within the time window will be downgraded. After this window time, the system will resume normal business processing logic.

[0033] goupLimitSet: the models retained after the model downgrade filtering on the terminal. If not configured, no downgrade will be performed;

[0034] channelGroupLimitMap: Channel downgrade filtering models configured by channel number;

[0035] partner-1: the models retained after the downgrade filtering of channel partner-1 models, no downgrade processing will be performed if no configuration is performed;

[0036] partner-2: The models retained after the downgrade filtering of channel partner-2 models. No downgrade processing will be performed if no configuration is performed.

[0037] Preferably, the above configuration sets a vehicle model restriction for the system and sets a system threshold. When the request volume exceeds the threshold, the requests from channel partner-1 and partner-2 within the system degradation time window will only estimate and return the corresponding vehicle models.

[0038] Preferably, the configurations in all vehicle model downgrade configuration files support dynamic modification and take effect in real time.

[0039] Preferably, the setting of the system threshold can be manually set according to different cities.

[0040] Compared with the prior art, the beneficial effects of the present invention are as follows: stress testing found that the estimated number of vehicle models is inversely proportional to the system throughput, the more estimated vehicle models, the lower the system throughput, and conversely, the fewer estimated vehicle models, the higher the system throughput; during holiday peaks, when the system reaches a traffic bottleneck, technical means are used to adjust the configuration of the estimated available vehicle models, only estimate commonly used vehicle models, and filter out uncommon vehicle models to save system resources, thereby increasing the throughput of a single machine; while increasing platform revenue, it can also improve the user experience of both drivers and passengers, indirectly improving the company's reputation. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] See also Figure 1The present invention provides a technical solution: a method for improving the estimated peak-period single-machine throughput. When it is found that a single machine is close to the system bottleneck, the big data algorithm model can be used to downgrade the configuration according to the vehicle model, filter out uncommon vehicle models, and only estimate the commonly used vehicle models. Ultimately, the system can provide normal services without capacity expansion during peak periods, allowing the platform to handle higher traffic.

[0044] The vehicle type filtering configuration file used in the present invention is configured as follows:

[0045] {

[0046] "groupLimit":{

[0047] "useLimit":true,

[0048] "sizeLimit":300,

[0049] "sizeLimitTimeout":100,

[0050] "goupLimitSet":["34","43","283"],

[0051] "channelGroupLimitMap":{

[0052] "partner-1":["34","43","283"],

[0053] "partner-2":["34","43","283"]

[0054] }

[0055] }

[0056] }

[0057] The above configuration description:

[0058] useLimit: whether to use the estimated vehicle type downgrade filter, true to use, false not to use.

[0059] sizeLimit: system initialization threshold. Downgrade occurs when this threshold is reached. The system automatically updates this threshold based on internal resource monitoring.

[0060] sizeLimitTimeout: System downgrade time window. All requests within the time window will be downgraded. After this window time, the system will resume normal business processing logic.

[0061] goupLimitSet: The models retained after the model downgrade filtering on the end. If not configured, no downgrade processing will be performed.

[0062] channelGroupLimitMap: Channel downgrade filtering models configured by channel number.

[0063] partner-1: The models retained after the downgrade filtering of channel partner-1 models. No downgrade processing will be performed if no configuration is performed.

[0064] The above configuration sets the system to use vehicle model restrictions and sets the system threshold to 300. When the request volume exceeds the threshold, the requests from channel partner-1 and partner-2 within 100 seconds will only estimate and return the three vehicle models 34, 43, and 283. The above configurations all support dynamic modification and take effect in real time.

[0065] The specific implementation plan is as follows:

[0066] S1: The estimation service configures downgraded models based on the common models provided by big data and configures the system initialization threshold based on the stress test results.

[0067] S2: The upstream makes multi-model estimation requests to the estimation service.

[0068] S3: The system checks whether the estimated service itself is in a degraded state.

[0069] S4: The estimation service obtains resource monitoring information in real time before processing the estimation, including system thread pool usage and CPU usage, and updates the system threshold in real time.

[0070] S5: The estimation service determines whether the system throughput reaches a threshold.

[0071] S6: If the estimated service throughput reaches the threshold, a synchronization lock is added to write the service degradation status. The service status is modified with the keyword volatile to ensure thread visibility. At the same time, a delayed task is started to update the service status to normal.

[0072] S7: If the estimated service is in a downgraded state, a comparison is made to see whether a vehicle model downgrade filter is configured and whether the estimated vehicle model includes a downgraded vehicle model.

[0073] S8: If vehicle type downgrade filtering is configured and the estimated vehicle type includes downgraded models, the estimation parameters are reorganized and only the downgraded models are estimated.

[0074] S9: The estimation service performs a normal estimation and returns.

[0075] S10: It is estimated that the service delay task will recover from the degraded state after a certain period of time.

[0076] 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 variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for improving the estimated peak-period single-machine throughput, characterized by: When a single machine is found to be close to the system bottleneck, the big data algorithm model can be used to downgrade the configuration according to the vehicle model, filter out uncommon models, and only estimate the commonly used models. Ultimately, the system can provide normal services without capacity expansion during peak hours, allowing the platform to handle higher traffic. The specific steps are as follows: S1: The estimation service configures downgraded models based on the common models provided by big data, and configures the system initialization threshold based on the stress test results; S2: The upstream makes a multi-car estimation request to the estimation service; S3: The system checks whether the estimated service itself is in a degraded state; S4: The estimation service obtains resource monitoring information in real time before processing the estimation, including system thread pool usage and CPU usage, and updates the system threshold in real time; S5: The estimation service determines whether the system throughput reaches the threshold; S6: If the estimated service throughput reaches the threshold, a synchronization lock is added to write the service degradation status. The service status is modified with the keyword volatile to ensure thread visibility. At the same time, a delayed task is started to update the service status to normal. S7: If the estimated service is in a downgraded state, check whether the vehicle model downgrade filter is configured and whether the estimated vehicle model includes a downgraded vehicle model; S8: If vehicle type downgrade filtering is configured and the estimated vehicle type includes downgraded models, the estimation parameters are reorganized to estimate only the downgraded models; S9: The estimation service performs normal estimation and returns; S10: It is estimated that the service delay task will recover from the degraded state after a certain period of time.

2. The method for improving the estimated peak-period stand-alone throughput according to claim 1, characterized in that: The vehicle model downgrade configuration file used is configured as follows: { "groupLimit":{ "useLimit":, "sizeLimit":, "sizeLimitTimeout":, "goupLimitSet":[], "channelGroupLimitMap":{ "partner-1":[], "partner-2":[] } } } The above configuration description: useLimit: whether to use the estimated vehicle model downgrade filtering, including true to use, false not to use; sizeLimit: system initialization threshold. Downgrade occurs when this threshold is reached. The system automatically updates this threshold based on internal resource monitoring. sizeLimitTimeout: system downgrade time window. All requests within the time window will be downgraded. After this window time, the system will resume normal business processing logic. goupLimitSet: the models retained after the model downgrade filtering on the terminal. If not configured, no downgrade will be performed; channelGroupLimitMap: Channel downgrade filtering models configured by channel number; partner-1: the models retained after the downgrade filtering of channel partner-1 models, no downgrade processing will be performed if no configuration is performed; partner-2: The models retained after the downgrade filtering of channel partner-2 models. No downgrade processing will be performed if no configuration is performed.

3. The method for improving the estimated peak-period stand-alone throughput according to claim 2, characterized in that: The above configuration sets the vehicle model restriction for the system and sets the system threshold. When the request volume exceeds the threshold, the requests from channel partner-1 and partner-2 within the system degradation time window will only estimate and return the corresponding vehicle models.

4. The method for improving the estimated peak-period stand-alone throughput according to claim 2, characterized in that: The configurations in all vehicle model downgrade configuration files support dynamic modification and take effect in real time.

5. The method for improving the estimated peak-period stand-alone throughput according to claim 3, characterized in that: The system threshold can be set manually according to different cities.

Images