Abstract: A method transcodes a media program to produce a constant video quality transcoded version of the media program. The constant video quality transcoded version is transcoded by a transcoder operating in a first mode that targets video quality. A portion of the constant video quality transcoded version that fails to satisfy a constraint is determined. The method sets a transcoding parameter based on the portion failing to satisfy the constraint. A portion of the media program corresponding to the portion of the constant video quality transcoded version is transcoded according to the transcoding parameter to produce a constant bit rate version of the portion. The constant bit rate version is transcoded by the transcoder operating in a second mode that targets a bit rate. The method then substitutes the constant bit rate version for the portion of constant video quality transcoded version in generating a transcoded media program.

Abstract: A video delivery service may display different content on an interface, such as ad campaign content, media program campaign content, and recommendation content. For example, the video delivery service may communicate with an ad engine to determine ad campaign content that is available and ad campaign scores to rate the ad campaign content. The video delivery service may also communicate with a media program campaign engine to determine any media program campaign content for the user and media program campaign scores to rate the media program campaign content. Then, the video delivery service may communicate with a recommendation engine to rate the relevance of the ad campaign content or media program campaign content to the user. After receiving the ad campaign content, media program campaign content, and recommendation content, the video delivery service determines which of the ad campaign content, media program campaign content, and recommendation content to display in the interface.

Abstract: In some embodiments, a method may receive a seekable range position for the live edge of a live media program and a current playback time. A distance between the position and the current playback time is calculated and compared to three or more feature ranges. A first range is where the user is considered watching live, a second range is where the user is considered not watching live, and a third range is where the user state is not changed. The method enables a first feature that displays a first control indicating that the media program is in a live state for a first range. The method enables a second feature that displays a second control to allow seeking to the live edge for a third range. The method does not change whichever of the first feature or the second feature this is currently enabled for a second range.

Abstract: In one embodiment, a method encodes a first set of segments of a media file based on a first bitrate range to generate a first set of encoded segments. It is then determined when an encoded segment for the first bitrate range includes a bitrate that satisfies a second bitrate range. When the encoded segment for the first bitrate range includes the bitrate that satisfies the second bitrate range, the method does not encode a segment in the first set of segments that corresponds to the encoded segment when the media file is encoded based on the second bitrate range to generate a second set of encoded segments for the second bitrate range. A first playlist for the first bitrate range is identified and a second playlist for the second bitrate range is identified. The second playlist also includes the encoded segment that was encoded for the first bitrate range.

Abstract: In one embodiment, a method generates a plurality of sub-relevance tables including a first set of relevance values between media programs. Each table models relevance values for a single feature in a plurality of features. Labeling results are received that include a second set of relevance values between the media programs. The method combines the sub-relevance tables into a single relevance table that includes a third set of relevance values between the media programs for the plurality of features. The combining generates weights for each of the sub-relevance tables based on the second set of relevance values for the labeling results and the first set of relevance values of the sub-relevance tables that are used to generate the third set of relevance values. A recommendation is provided to a user using the third set of relevance values from the single relevance table and a characteristic of the user.

Abstract: Particular embodiments execute tasks to measure performance in a computing system. The method uses a master computing device and helper client computing devices. The helper client computing devices may be situated in a pool where the helper client computing devices are available to help a master computing device to perform a task. When the master computing device wants to perform a task, the master computing device may send a message to the pool requesting help with a task. Helper client computing devices can respond to the message when the helper clients are available to join in groups to process tasks. Once the master computing device configures a group with helper client computing devices that responded to the message, the master computing device and the helper client computing devices perform the task together.

Abstract: In one embodiment, during execution of a current workload being processed by the kernel, the method searches policy storage for a similar workload that has been previously optimized, wherein the sensor is used to compare the current workload to workloads in the policy storage. When the similar workload is found, the method optimizes different parameter values in a parameter value range specified in the policy specification for the parameter while the current workload is being processed by the kernel, wherein a parameter setting for the similar workload is used in the optimizing. When the similar workload is not found, the method optimizes different parameter values based on the parameter value range specified in the policy specification for the parameter while the current workload is being processed by the kernel. The method then evaluates the optimizing based on the optimization target.

Abstract: In one embodiment, a method sends videos to a user that is using a video delivery service. The method receives user behavior that includes actions taken by the user on the video delivery service. The method inputs the user behavior into a first predictor to generate a set of actions for a set of entities. Also, the method inputs the set of actions for the set of entities, a real-time context, and the user behavior into a second predictor to generate probabilities for the set of actions for the set of entities. A probability for an action indicates the probability the user would select that action for an entity when compared against other actions in the set of actions for the set of entities. An action feed is selected based on the ranking and dynamically output to a client while the user is using the video delivery service.

Abstract: In one embodiment, a method receives interest indications for entities, entitlements, and location information that are indexed by user profiles in databases. The interest indications for the entities, the entitlements, and the location information are transformed from being indexed by the user profiles to indexing the entities and indexing entitlement and location information as availability pairs in an index and associating user profiles in the user profiles as entries for the index. The method receives a notification of a live programming event before the event starts and uses the notification to determine an entity of a media program and an availability pair. A second database is queried using the entity and the availability pair to determine a set of user profiles associated with the entity and the availability pair. Then, an action is performed for at least a portion of the set of user profiles before the live programming event occurs.

Abstract: Particular embodiments allow a content delivery service to dynamically adjust how the service allocates content delivery traffic to multiple content delivery networks (CDNs). In one embodiment, the content delivery service may service enough requests for media programs that multiple CDNs are used. This allows the content delivery service to allocate traffic to different CDNs. Instead of solely relying on the CDN to provide quality of service solutions, the content delivery service may configure client devices to send information to it that allows the content delivery service to analyze the media program playback quality of service. Then, the content delivery service can adjust the allocation for the CDNs if a problem is detected at a specific CDN.

Abstract: Particular embodiments use a process that can invalidate an elected leader using an invalidated leader value. The availability and use of the invalidated leader value can avoid the requirement of performing a new election round to elect a new leader. When one node of the group detects that there may be a fault with respect to the leader of the system, the node can start the process to establish a new leader autonomously. First, the node can invalidate the leader. Then, the node attempts to propose a new leader. If a quorum is received, then the proposed leader may be elected as the new leader. By invalidating the old leader, the node can ensure that the old leader cannot be elected the new leader once the quorum is received for the new leader.

Abstract: Particular embodiments provide a test system that allows the testing of different versions. The test system uses a repository of features that are included in the multiple versions of the application. When new features are added to the application, they may be added to the repository. When a test is performed, an application blueprint is used to build an application model using features from the repository. Then, a test case for the version can be run using the application model. By maintaining the repository of features and then dynamically building the application model for the version of the application, the testing of different versions of the application can be performed.

Abstract: In one embodiment, a method receives information describing programs for a channel over multiple time slots. Audience definitions are generated for the programs defining parameters for selecting a respective source stream for delivery to users. A bundle is stored for each audience definition. A request is received for the channel from a user. The method then generates a viewer definition for the user based on a current status of the user, wherein the viewer definition includes user information for the user. In response to receiving the request, a manifest is constructed for the channel using a set of bundles that includes audience definitions that match the viewer definition. Only one bundle per time slot is selected for the manifest. Portions of the manifest are provided to a client device to allow the client device to request content from the channel.

Abstract: In one embodiment, a method plays, on an interface, a media program in a media player operating in a first mode. A position change of the media player on the interface is determined while the media program is being played in the media player to cause the media player to move to an out of bounds position. The method detects the media player moving to the out of bounds position and upon detecting, automatically performs switching from the first mode to a second mode based on the media player moving to the out of bounds position where the second mode restricts movement of the media player. Also, the method transfers the media player to a new position that is not in the out of bounds position where the media player continues to play the media program in the new position.

Abstract: A method receives a candidate set of recommendations for video entities on a video delivery service in response to receiving a request to generate a page of an interface. A number for each recommendation is generated that represents a relevance rating of the respective recommendation minus a similarity rating between the respective recommendation and recommendations from the candidate set of recommendations that are added to a subset of recommendations. A recommendation is added to the subset of recommendations that has a maximum probability of being relevant to the user and diverse from the recommendations in the subset of recommendations based on the number. The method then updates the number for recommendations in the candidate set of recommendations based on adding the recommendation to the subset of recommendations. This process is iteratively performed and the subset of recommendations in the page of the interface is provided to a client device.

Abstract: Particular embodiments generate a database table that keeps track of authorization package combinations for videos being offered by a video delivery service. The table may allow the video delivery service to keep track of how much content is available for a specific authorization package combination that a user might have access to In one embodiment, the video delivery service does not pre-generate every single possible authorization package combination on a per-show basis. Rather, only a subset of the authorization package combinations is generated for each show. Then, the video delivery service uses a query to the database table that can calculate which content a user is authorized to access from the subset of authorization package combinations that were generated.