Selective content pre-warming in content delivery networks based on user actions and content categorizations
View Patent ↗Systems and methods are provided for streaming content over the Internet via a CDN to an end user system. Requests from end user systems for streaming content objects being handled by the CDN are tracked to identify, for each request the tracked requests, a first content object being requested. For each request of the tracked requests, a category of the requested first content object is identified. The identified category is stored. Aggregated data indicating an amount of requests being for content objects with a specific category is generated. A prediction is made that a second content object will be requested in the future based on the aggregated data and a category of the second content object. Access to the second content object is improved such that the second content object is set to be provided faster in response to a request for the second content object than would otherwise occur.
1. A content delivery network (CDN) for streaming content over the Internet to end user systems, the CDN comprising:
a content request interface that receives requests, using one or more processors, for streaming a plurality of content objects to end user systems;
an edge server selection function that assigns, using one or more processors, each request to an edge server for streaming the plurality of content objects to the end user systems;
a streaming function that streams, using one or more processors, the plurality of content objects to the end user systems; and
a content-request function, wherein the content-request function, using one or more processors:
identifies, for each request received by the content request interface, a category for each of the plurality of content objects, wherein the category for each of the plurality of content objects is identified based on an analysis of the subject matter of each of the plurality of content objects,
stores the category for each of the plurality of content objects,
monitors, for each request received, a first portion of each of the plurality of content objects that was streamed to a requesting end user system, wherein the first portion is less than an entire content object,
generates aggregated data to determine a playback time, wherein:
the playback time corresponds to a specific category of content objects of the plurality of content objects; and
the playback time is equal to a statistical usage pattern of first portions of the plurality of content objects that belong to the specific category,
predicts a second portion of a second content object, wherein:
the second content object is not part of the plurality of content objects,
the second content object is identified to be part of the specific category, and
the second portion has a playback length equal to the playback time based on the second content object being identified to be part of the specific category, and
causes access to the second portion of the second content object to be improved such that the second portion of the second content object is set to be provided faster in response to a request for the second content object than would otherwise occur.
2. The CDN for streaming content over the Internet to the end user system as recited in claim 1 , wherein causing access to the second content object to be improved includes causing the second content object to be cached at an edge server.
3. The CDN for streaming content over the Internet to the end user system as recited in claim 1 , wherein, for each request received by the content request interface, the category of the requested content object is identified based on metadata of the requested content object.
4. The CDN for streaming content over the Internet to the end user system as recited in claim 1 , wherein the content-request function further:
monitors, for each request received by the content request interface, what portion of the request was streamed to the requesting end user system, and
predicts that a portion of the second content object will be provided in response to future requests,
wherein causing access to the second content object to be improved comprises causing access to the portion of the second content object to be improved.
5. The CDN for streaming content over the Internet to the end user system as recited in claim 1 , wherein predicting that the second content object will be requested in the future comprises predicting that a particular user will request the second content object.
6. The CDN for streaming content over the Internet to the end user system as recited in claim 5 , wherein the content-request function further estimates an edge server that will be serving the user in the future, and wherein causing access to the second content object to be improved includes causing the second content object to be cached at the edge server.
7. The CDN for streaming content over the Internet to the end user system as recited in claim 1 , wherein predicting that the second content object will be requested comprises predicting that the second content object will be requested more than a threshold number of times.
8. The CDN for streaming content over the Internet to the end user system as recited in claim 1 , wherein the generated aggregated data is indicative of an amount of requests assigned to a specific point of presence (POP) of the CDN being for content objects with the specific category, and wherein the prediction comprises a prediction that the specific POP will receive a future request for the second content object.
9. A method for streaming content over the Internet via a CDN to end user systems, the method comprising:
tracking requests from end user systems for streaming content objects being handled by the CDN to identify a plurality of content objects being requested;
identifying a category for each of the plurality of content objects, wherein the category for each of the plurality of content objects is identified based on an analysis of the subject matter of each of the plurality of content objects;
storing the category for each of the plurality of content objects;
monitoring, for each request received, a first portion of each of the plurality of content objects that was streamed to a requesting end user system, wherein the first portion is less than an entire content object;
generating aggregated data to determine a playback time, wherein:
the playback time corresponds to a specific category of content objects of the plurality of content objects; and
the playback time is equal to a statistical usage pattern of first portions of the plurality of content objects that belong to the specific category;
predicting that a second portion of a second content object, wherein:
the second content object is not part of the plurality of content objects;
the second content object is identified to be part of the specific category; and
the second portion has a playback length equal to the playback time based on the second content object being identified to be part of the specific category; and
causing access to the second portion of the second content object to be improved such that the second portion of the second content object is set to be provided faster in response to a request for the second content object than would otherwise occur.
10. The method for streaming content over the Internet to the end user system as recited in claim 9 , wherein causing access to the second content object to be improved includes causing the second content object to be cached at an edge server in the CDN.
11. The method for streaming content over the Internet to the end user system as recited in claim 9 , wherein, for each request received by the content request interface, the category of the requested content object is identified based on metadata of the requested content object.
12. The method for streaming content over the Internet to the end user system as recited in claim 9 , further comprising:
monitoring, for each request of the tracked requests, what portion of the request was streamed to the requesting end user system; and
predicting that a portion of the second content object will be provided in response to future requests,
wherein causing access to the second content object to be improved comprises causing access to the portion of the second content object to be improved.
13. The method for streaming content over the Internet to the end user system as recited in claim 9 , wherein predicting that the second content object will be requested in the future comprises predicting that a particular user will request the second content object.
14. The method for streaming content over the Internet to the end user system as recited in claim 13 , further comprising estimating an edge server of the CDN that will be serve the user in the future, wherein causing access to the second content object to be improved includes causing the second content object to be cached at the edge server.
15. The method for streaming content over the Internet to the end user system as recited in claim 9 , wherein predicting that the second content object will be requested comprises predicting that the second content object will be requested more than a threshold number of times.
16. The method for streaming content over the Internet to the end user system as recited in claim 9 , wherein the generated aggregated data is indicative of an amount of requests assigned to a specific point of presence (POP) of the CDN being for content objects with the specific category, and wherein the prediction comprises a prediction that the specific POP will receive a future request for the second content object.
17. A system for streaming content over the Internet to end user systems, the system comprising:
one or more processors; and
one or memories coupled with said one or more processors, wherein the one or more processors and one or more memories are configured to:
track requests from end user systems for streaming a plurality of content objects;
identify a category for each of the plurality of content objects, wherein the category for each of the plurality of content objects identified based on an analysis of the subject matter of each of the plurality of content objects;
store the category for each of the plurality of content objects;
monitor, for each request received, a first portion of each of the plurality of content objects that was streamed to a requesting end user system, wherein the first portion is less than an entire content object;
generate aggregated data to determine a playback time, wherein:
the playback time corresponds to a specific category of content objects of the plurality of content objects; and
the playback time is equal to a statistical usage pattern of first portions of the plurality of content objects that belong to the specific category;
predict that a second portion of a second content object, wherein:
the second content object is not part of the plurality of content objects;
the second content object is identified to be part of the specific category; and
the second portion has a playback length equal to the playback time based on the second content object being identified to be part of the specific category; and
cause access to the second portion of the second content object to be improved such that the second portion of the second content object is set to be provided faster in response to a request for the second content object than would otherwise occur.
18. The system for streaming content over the Internet to the end user system as recited in claim 17 , wherein causing access to the second content object to be improved includes causing the second content object to be cached.
19. The system for streaming content over the Internet to the end user system as recited in claim 17 , wherein, for each request received by the content request interface, the category of the requested content object is identified based on metadata of the requested content object.
20. The system for streaming content over the Internet to the end user system as recited in claim 17 , wherein the one or more processors and one or more memories are further configured to:
monitor, for each request of the tracked requests, what portion of the request was streamed to the requesting end user system; and
predict that a portion of the second content object will be provided in response to future requests,
wherein causing access to the second content object to be improved comprises causing access to the portion of the second content object to be improved.