Resource allocation optimization for cloud-based video processing

• US 9,407,944 B1
• Filed: 05/08/2015
• Issued: 08/02/2016
• Est. Priority Date: 05/08/2015
• Status: Active Grant

• Current Assignee: iStreamPlanet Co. LLC (Warner Bros. Discovery, Inc.)
• Original Assignee: iStreamPlanet
• Inventors: Galdy, Alessio, Juteau, Eric, Davis, Jeffrey Alan, VanAntwerp, Mark Daniel
• Primary Examiner(s): Asanbayev, Oleg
• Application Number:
  US14/708,128
• Time in Patent Office: 452 Days
• US Class Current: 1/1
• CPC Class Codes: H04N 21/2181 comprising remotely distributed storage units, e.g. when movies are replicated over a plurality of video servers distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS], H04L67/1097
• CPC Class Codes: H04N 21/2225 Local VOD servers
• CPC Class Codes: H04N 21/23103 using load balancing strategies, e.g. by placing or distributing content on different disks, different memories or different servers storage management G06F3/0604; allocation of resources considering the load in multiprogramming arrangements G06F9/505; techniques for rebalancing the load in a distributed system G06F9/5083; access to distributed or replicated servers, e.g. load balancing, in data networks H04L67/1001
• CPC Class Codes: H04N 21/2343 involving reformatting operations of video signals for distribution or compliance with end-user requests or end-user device requirements video transcoding H04N19/40; media packet handling at the source H04L65/762
• CPC Class Codes: H04N 21/23655 Statistical multiplexing, e.g. by controlling the encoder to alter its bitrate to optimize the bandwidth utilization
• CPC Class Codes: H04N 21/2368 Multiplexing of audio and video streams
• CPC Class Codes: H04N 21/2385 Channel allocation H04N21/266 takes precedence; Bandwidth allocation H04N21/24 takes precedence ; allocation of channels according to the instantaneous demands of the users in time-division multiplex systems H04J3/1682; admission control, resource allocation in open networks H04L12/5692; arrangements for maintenance or administration in data switching networks involving bandwidth and capacity management H04L41/0896; negotiating bandwidth in wireless networks H04W28/16
• CPC Class Codes: H04N 21/2402 Monitoring of the downstream path of the transmission network, e.g. bandwidth available traffic monitoring in data switching networks H04L43/00; monitoring data switching networks utilization H04L43/0876
• CPC Class Codes: H04N 21/2405 Monitoring of the internal components or processes of the server, e.g. server load allocation of resources in multiprogramming arrangements G06F9/50; performance measurement of computer activity G06F11/34
• CPC Class Codes: H04N 21/2743 Video hosting of uploaded data from client
• CPC Class Codes: H04N 21/64322 IP
• CPC Class Codes: H04N 21/6437 Real-time Transport Protocol [RTP]
• CPC Class Codes: H04N 21/812 involving advertisement data advertising per se G06Q30/02

Abstract

A method executed by an electronic device to optimize resource utilization while processing media workflows by a video streaming platform is disclosed. In one embodiment, a request to initiate a set of training sessions is received, and for each training session, a set of training workflows to be processed is initiated and a task graph for each training workflow is created. Then a worker of the video streaming platform is assigned to each training session. The raw performance data from each assigned worker is then collected, and a platform performance profile is generated based on the raw performance data from the each assigned worker. The platform performance profile is used to predict resource requirements of media workflows to be processed by the video streaming platform. A system to perform the method and a storage medium storing instructions to perform the method are disclosed too.

Claims

1. A method executed by an electronic device implementing a video streaming platform, the method to optimize resource utilization while processing media workflows by the video streaming platform, the method comprising:

receiving a request to initiate a set of training sessions, the request including a set of parameters describing characteristics of each training session in the set of training sessions;

for each training session in the set of training sessions, initiating a set of training workflows to be processed by the video streaming platform;

creating a task graph for each training workflow in the set of training workflows, the task graph being a directed acyclic graph of tasks with each node of the task graph representing a media processing task and each edge of the task graph representing a data flow across two tasks and a corresponding input and output of each task;

assigning a worker from a set of workers of the video streaming platform to each training session;

collecting raw performance data from each assigned worker after each set of training workflows is processed;

generating a platform performance profile of the video streaming platform based on the raw performance data from the each assigned worker, wherein the platform performance profile is used to predict resource requirements of media workflows to be processed by the video streaming platform;

receiving a request to create a media workflow for a video source;

predicting a set of resource requirements necessary to process the media workflow based on the platform performance profile and characteristics of the video source, wherein the predicting of the set of resource requirements includes creating a task graph for the media workflow, and predicting the set of resource requirements for each task of the task graph based on the platform performance profile and the characteristics of the video source; and

selecting one or more workers based on the set of predicted resource requirements and current resource availability of workers in the video streaming platform.

2. The method of claim 1, wherein the set of parameters includes values for characteristics of the set of training sessions including video source codec, output audio/video transcoding, and publishing operations.

3. The method of claim 1, wherein the platform performance profile enables predictive modeling of resource requirements for specific task types in a media workflow.

4. The method of claim 1, wherein generating the platform performance profile comprises performing a linear regression on the raw performance data.

5. The method of claim 4, wherein the linear regression comprises generating a set of coefficients from the raw performance data from the worker.

6. The method of claim 1, wherein the raw performance data includes metrics on a processor usage.

7. The method of claim 6, wherein the processor usage includes a usage of a graphic processing unit.

8. An electronic device to serve as an orchestrator cluster of a video streaming platform, the electronic device comprising:

a processor; and

a non-transitory machine-readable storage medium coupled to the processor, the non-transitory machine-readable storage medium containing operations executable by the processor, wherein the electronic device is operative to receive a request to initiate a set of training sessions, the request including a set of parameters describing characteristics of each training session in the set of training sessions, for each training session in the set of training sessions, to initiate a set of training workflows to be processed by the video streaming platform, to create a task graph for each training workflow in the set of training workflows, the task graph being a directed acyclic graph of tasks with each node of the task graph representing a media processing task and each edge of the task graph representing a data flow across two tasks and a corresponding input and output of each task, to assign a worker from a set of workers of the video streaming platform to each training session, to collect raw performance data from each assigned worker after each set of training workflows is processed, to generate a platform performance profile of the video streaming platform based on the raw performance data from the each assigned worker, wherein the platform performance profile is used to predict resource requirements of media workflows to be processed by the video streaming platform, to receive a request to create a media workflow for a video source, to predict a set of resource requirements necessary to process the media workflow based on the platform performance profile and characteristics of the video source, wherein prediction of the set of resource requirements includes creating a task graph for the media workflow, and predicting the set of resource requirements for each task of the task graph based on the platform performance profile and the characteristics of the video source, and to select one or more workers based on the set of predicted resource requirements and current resource availability of workers in the video streaming platform.

9. The electronic device of claim 8, wherein the generation of the platform performance profile comprises performing a linear regression on the raw performance data.

10. The electronic device of claim 9, wherein the linear regression comprises generating a set of coefficients from the raw performance from the worker.

11. The electronic device of claim 8, wherein the raw performance data includes metrics on a processor usage.

12. A non-transitory machine-readable storage medium having instructions stored therein, which when executed by a processor, cause the processor to perform operations in an electronic device implementing a video streaming platform, wherein the operations optimize resource utilization while processing media workflows by the video streaming platform, the operations comprising:

receiving a request to initiate a set of training sessions, the request including a set of parameters describing characteristics of each training session in the set of training sessions;

for each training session in the set of training sessions, initiating a set of training workflows to be processed by the video streaming platform;

creating a task graph for each training workflow in the set of training workflows, the task graph being a directed acyclic graph of tasks with each node of the task graph representing a media processing task and each edge of the task graph representing a data flow across two tasks and a corresponding input and output of each task;

assigning a worker from a set of workers of the video streaming platform to each training session;

collecting raw performance data from each assigned worker after each set of training workflows is processed;

generating a platform performance profile of the video streaming platform based on the raw performance data from the each assigned worker, wherein the platform performance profile is used to predict resource requirements of media workflows to be processed by the video streaming platform;

receiving a request to create a media workflow for a video source;

predicting a set of resource requirements necessary to process the media workflow based on the platform performance profile and characteristics of the video source, wherein the predicting of the set of resource requirements includes creating a task graph for the media workflow, and predicting the set of resource requirements for each task of the task graph based on the platform performance profile and the characteristics of the video source; and

selecting one or more workers based on the set of predicted resource requirements and current resource availability of workers in the video streaming platform.

13. The storage medium of claim 12, wherein the platform performance profile enables predictive modeling of resource requirements for specific task types in a media workflow.

14. The storage medium of claim 12, wherein generating the platform performance profile comprises performing a linear regression on the raw performance data.

15. The storage medium of claim 12, wherein the raw performance data includes metrics on a processor usage.