Adaptive concurrency control for media pipeline

1 . A method of providing media data streams, comprising:

receiving, by a computing system, a network alert associated with a cloud network configured to provide a plurality of media data streams to user devices and comprising one or more network components, the network alert comprising request data associated with one or more requests received from respective user devices for at least one of the plurality of media data streams;

determining, by the computing system, a network status of the cloud network, the network status indicating a network load of the cloud network;

determining, by the computing system, that the network load exceeds a traffic threshold of the cloud network;

providing, by the computing system, at least one of the request data and the network status to a machine learning model, the machine learning model configured to assign a priority score of each of the plurality of media data streams based on the request data and/or the network status;

determining, by the computing system and using the priority scores of the machine learning model, a high-priority media data stream of the plurality of media data streams and a second media data stream of the plurality of media data stream;

determining, by a throttling module of the computing system, a first rate for the high-priority media data stream and a second rate for the second media data stream, based at least in part on the network load;

providing, by the computing system, first instructions to at least one of the one or more network components indicating that the high-priority media data stream of the plurality of media data streams is to be provided to a first user device according to a request for the high-priority media data stream at the first rate;

providing, by the computing system, second instructions to at least one of the one or more network components indicating that the second media data stream of the plurality of media data streams is to be provided to a second user device corresponding to a request for the second media data stream at the second rate, such that the high-priority media data stream and the second media data stream are provided to the first and second user devices, respectively, and the network load is below the traffic threshold;

receiving, by the computing system and from the second user device, a retry request for the second media data stream indicating the second media data stream was previously requested but not fulfilled;

modifying, by the machine learning model and based at least in part on the priority scores, a retry schedule; and

transmitting, by the computing system and based at least in part on the priority score, third instructions to the second user device to transmit subsequent retry requests to the cloud network according to the retry schedule.

2 . The method of claim 1 , further comprising:

determining, by the computing system, that the network load exceeds a traffic threshold of the cloud network; and

providing, by the computing system, instructions to at least one network component of the one or more network components to instantiate a new network component such that the traffic threshold of the network is increased above the network load indicated in the network status.

3 . The method of claim 1 , wherein the network alert comprises predicted request data generated by the machine learning model and the network status comprises a predicted network load generated by the machine learning model.

4 . The method of claim 3 , wherein the predicted request data and the predicted network load are generated by the machine learning model and based on one or more characteristics of the respective user devices.

5 . The method of claim 3 , wherein the predicted request data and the predicted network load are generated by the machine learning model and based metadata associated with the plurality of media data streams.

6 . The method of claim 1 , wherein the machine learning model is trained using at least one of one current event data, future event data, and data associated with historical media data streams.

7 . The method of claim 1 , wherein the priority score of each of the plurality of media data streams is determined in part based on a characteristic of the respective user devices.

8 . The method of claim 1 , wherein the retry schedule is based at least in part on the network load.

9 . A system for providing media data streams, comprising:

a network monitor;

a throttling module;

a machine learning model;

one or more processors; and

a non-transitory-computer readable medium comprising instructions that, when executed by the one or more processors, cause the system to perform operations to:

receive, by the network monitor, a network alert associated with a cloud network configured to provide a plurality of media data streams to user devices and comprising one or more network components, the network alert comprising request data associated with one or more requests received from respective user devices for at least one of the plurality of media data streams;

determine, by the network monitor, a network status of the cloud network, the network status indicating a network load of the cloud network;

determine, by the network monitor, that the network load exceeds a traffic threshold of the cloud network;

provide, by the network monitor, at least one of the request data and the network status to the machine learning model, the machine learning model configured to assign a priority score of each of the plurality of media data streams based on the request data and/or the network status;

determine, by the system and using the priority scores of the machine learning model, a high-priority media data stream of the plurality of media data streams and a second media data stream of the plurality of media data streams;

determine, by the throttling module of the system, a first rate for the high-priority media data stream and a second rate for the second media data stream, based at least in part on the network load;

provide, by the throttling module, instructions to at least one of the one or more network components indicating that the high-priority media data stream of the plurality of media data streams is to be provided to a first user device according to a request for the high-priority media data stream at a first rate;

provide, by the throttling module, instructions to at least one of the one or more network components indicating that the second media data stream of the plurality of media data streams is to be provided to a second user device corresponding to a request for the second media data stream at a second rate, such that the high-priority media data stream and the second media data stream are provided to the first and second user devices, respectively, and the network load is below the traffic threshold;

receive, by the system and from the second user device, a retry request for the second media data stream indicating the second media data stream was previously requested but not fulfilled;

modify, by the machine learning model and based at least in part on the priority scores, a retry schedule; and

transmit, by the system and based at least in part on the priority score, third instructions to the second user device to transmit subsequent retry requests to the cloud network according to the retry schedule.

10 . The system of claim 9 , wherein a portion of the request data comprises a retry request, and based at least in part on the network status, and the instructions further cause the system to:

determine a user device corresponding to the retry request; and

transmit, by the throttling module, instructions to the user device associated with the retry request to send subsequent retry requests to cloud network according to a retry schedule.

11 . The system of claim 9 , wherein the instructions further cause the system to:

determine, by the network monitor, that the network load exceeds a traffic threshold of the cloud network; and

provide, by the network monitor, instructions to at least one network component of the one or more network components to instantiate a new network component such that the traffic threshold of the network is increased above the network load indicated in the network status.

12 . The system of claim 9 , wherein the network alert comprises predicted request data generated by the machine learning model and the network status comprises a predicted network load generated by the machine learning model.

13 . The system of claim 12 , wherein the predicted request data is generated, at least in part by a second machine learning model and the predicted network load is generated, at least in part by a third machine learning model.

14 . A system comprising:

one or more processors; and

a non-transitory-computer readable medium comprising instructions that, when executed by the one or more processors, cause the system to perform operations to:

receive a network alert associated with a cloud network configured to provide a plurality of media data streams to user devices and comprising one or more network components, the network alert comprising request data associated with one or more requests received from respective user devices for at least one of the plurality of media data streams;

determine a network status of the cloud network, the network status indicating a network load of the cloud network;

provide at least one of the request data and the network status to a machine learning model, the machine learning model configured to assign a priority score of each of the plurality of media data streams based on the request data and/or the network status;

determine, using the priority scores of the machine learning model, a high-priority media data stream of the plurality of media data streams and a second media data stream of the plurality of media data streams;

provide instructions to at least one of the one or more network components indicating that the high-priority media data stream of the plurality of media data streams is to be provided to a first user device according to a request for the high-priority media data stream at a first rate, the first rate determined based at least in part on the network load;

provide instructions to at least one of the one or more network components indicating that the second media data stream of the plurality of media data streams is to be provided to a second user device corresponding to a request for the second media data stream at a second rate, the second rate based at least in part on the network load;

receive, by the system and from the second user device, a retry request for the second media data stream indicating the second media data stream was previously requested but not fulfilled;

modify, by the machine learning model and based at least in part on the priority scores, a retry schedule; and

transmit, by the system and based at least in part on the priority score, third instructions to the second user device to transmit subsequent retry requests to the cloud network according to the retry schedule.

15 . The system of claim 14 , wherein a portion of the request data comprises a retry request, and based at least in part on the network status, and the instructions further cause the system to:

determine a user device corresponding to the retry request; and

transmit instructions to the user device associated with the retry request to send subsequent retry requests to the cloud network according to a retry schedule.

16 . The system of claim 14 , wherein the network alert comprises predicted request data generated by the machine learning model and the network status comprises a predicted network load generated by the machine learning model.

17 . The system of claim 14 , the priority score of each of the plurality of media data streams is determined in part based on a characteristic of the respective user devices.

18 . The method of claim 1 , wherein the retry schedule includes a number of retry requests, an interval between subsequent retry requests, or both the number of retry requests and the interval between subsequent retry requests.

19 . The method of claim 18 , wherein modifying the retry schedule comprises modifying a number of retry requests the user device is permitted to perform prior to halting subsequent retry requests.

20 . The method of claim 18 , wherein modifying the retry schedule comprises modifying an interval between subsequent retry requests to be sent by the user device.