Abstract: Some examples include detecting that a source computing device is in proximity to a target computing device. For instance, a user credential of the source computing device may be received from the source computing device. Further, a copy of system data of the source computing device may be retrieved from a cloud storage service. A system state of the source computing device may be restored on the target computing device by using the user credential received from the source computing device and the copy of the system data of the source computing device retrieved from the cloud storage. In some cases, the user credential may expire on the target computing device after a predetermined system event in the target computing device.

Abstract: Some examples include optimizing system performance by generating a version vector that includes a plurality of attributes associated with a computing device. For instance, the plurality of attributes associated with the computing device may include a device type of the computing device. The plurality of attributes from the version vector may be analyzed to determine a usage pattern of a user of the computing device. In addition, a configuration of the computing device may be adjusted based on the usage pattern. In some cases, adjusting of the configuration of the computing device may include adjusting, based on the usage pattern, one or more of a power consumption of the computing device or a CPU running time of the computing device.

Abstract: Some examples include performing a priority synchronization of a computing device to download a selected set of data files to the computing device from a distributed backup system. Data files are downloaded to a computing device by synchronizing with a server of the distributed backup system. A priority sync downloads a subset of the data files as priority files, rather than all of the data files of the user, thereby minimizing the computing resources consumed to download the data files. The priority sync can select the priority files based on various criteria, including an access pattern of the data files, attributes of the data files, or attributes of the computing device. The priority sync may also download non-priority files. The priority files may be downloaded in their original format and the non-priority files may be downloaded in a downgraded format.

Abstract: Some examples include backing up and retrieving data in a distributed backup system. For instance, a server may receive a request to backup multiple data files of a user to multiple computing devices associated with the user. Each of the multiple computing devices may be associated with one or more data backup policies for storing a set of data files. The server may identify, for each of the data files, the one or more data backup policies with which the corresponding data file matches and a corresponding set of the multiple computing devices associated with the one or more of the data backup policies. Further, the server may transmit each of the data files to the corresponding set of computing devices.

Abstract: Various embodiments generally relate to decentralized operating systems spanning multiple devices. More specifically, some embodiments relate to continuous and concurrent device experiences across the multiple devices. For example, a user may be streaming a movie through a service provided via a smart television. When the user's phone detects an incoming video call, the television can recognize that a video call is coming in and ask the user if the user wants to answer the call. If the user responds in the affirmative, the television can pause the streaming video from the service provide allowing the user to answer their phone. In some cases, the television can be used as the display screen to display the video call while the local phone resources (e.g., video camera) can be used as the recording device.

Abstract: Some examples include one or more processors that optimize the storage capacity of a computing device of a user. For instance, the one or more processors may receive, from the computing device, an indication of a request to download a data file. In addition, the one or more processors may determine a device classification associated with the computing device and a user access pattern associated with the data file. Further, the one or more processors may perform a downsampling of the data file based on the device classification and the user access pattern in response to the request, and may transmit the downsampled data file to the computing device.

Abstract: Various embodiments are disclosed for enabling multiple computing devices having a decentralized operating system to effectively monitor and manage resources across the multiple devices. In some embodiments, the systems are able to identify a set of local resources available for utilization by an application running on the decentralized operating system. A resource message identifying the set of local resources can be generated using a resource discovery and utilization application programming interface. This information can then be sent to a requestor (e.g., an application running on the decentralized operating system) which can create a resource utilization plan.

Abstract: Some examples include optimizing system performance by generating a version vector that includes a plurality of attributes associated with a computing device. For instance, the plurality of attributes associated with the computing device may include a device type of the computing device. The plurality of attributes from the version vector may be analyzed to determine a usage pattern of a user of the computing device. In addition, a configuration of the computing device may be adjusted based on the usage pattern. In some cases, adjusting of the configuration of the computing device may include adjusting, based on the usage pattern, one or more of a power consumption of the computing device or a CPU running time of the computing device.

Abstract: Some examples include detecting that a source computing device is in proximity to a target computing device. For instance, a user credential of the source computing device may be received from the source computing device. Further, a copy of system data of the source computing device may be retrieved from a cloud storage service. A system state of the source computing device may be restored on the target computing device by using the user credential received from the source computing device and the copy of the system data of the source computing device retrieved from the cloud storage. In some cases, the user credential may expire on the target computing device after a predetermined system event in the target computing device.

Abstract: Some examples include backing up and restoring a game application state across multiple devices. A distributed system may run an instance of a game application at a first electronic device, determine a backup event that occurs in the first electronic device, wherein the backup event suggests a backup of application state data and the application state data represent an application state of the game application at the backup event, and transmit the application state data of the game application to a remote storage service in response to the backup event. A second electronic device may retrieve the application state data from the remote storage service and restore the game application state.

Abstract: Technology disclosed herein includes a method for system restoration between computing devices by a tapping mechanism. A first device detects (e.g., via NFC) a second device in proximity. The first device transmits a user credential of the first device to the second device. Either directly or via a cloud storage service, the first device instructs the second device to download a copy of system data of the first device from the cloud storage service, instead of from the first device. The second device is able to restore a system state of the first device on the second device, using the user credential and the system data.

Abstract: Some examples include caching data among multiple interconnected computing devices. As one example, a storage server may transmit a first data set to a first computing device, and may identifying a second computing device from a plurality of computing devices sharing a local area network with the first computing device. For instance, the second computing device may be selected from the plurality of computing devices based on battery levels of the plurality of computing devices. The storage server may identify a second data set as relevant to the first data set and may transmit the identified second data set to the second computing device. Additionally, the storage server may transmit an instruction to the first computing device indicating that the second computing device stores a data cache for the first computing device.

Abstract: Some examples include performing a priority synchronization of a computing device to download a selected set of data files to the computing device from a distributed backup system. Data files are downloaded to a computing device by synchronizing with a server of the distributed backup system. A priority sync downloads a subset of the data files as priority files, rather than all of the data files of the user, thereby minimizing the computing resources consumed to download the data files. The priority sync can select the priority files based on various criteria, including an access pattern of the data files, attributes of the data files, or attributes of the computing device. The priority sync may also download non-priority files. The priority files may be downloaded in their original format and the non-priority files may be downloaded in a downgraded format.

Abstract: Technology is disclosed for classification of data objects (e.g., files) in a distributed file system based on application creation and/or access information. In some embodiments, data objects of a user are stored across multiple computing devices associated with a user, e.g., mobile computing devices such as a tablet, a laptop, or a smartphone, and/or at server systems, e.g., cloud-based storage servers, in a distributed cloud-based storage system. The technology described tracks the applications that create or touch particular data objects in the distributed file system and classifies the data objects according to these applications resulting in various application classifications. For example, a classification engine can process data objects in the distributed file system to identify an application that created the data object and/or applications that have accessed the data object. The classification information can be saved by, for example, a cloud-based storage server.

Abstract: Technology is disclosed for performing a priority synchronization of a computing device to download a selected set of data files to the computing device from a distributed backup system. Data files are downloaded to a computing device by synchronizing the computing device with a server of the distributed backup system. A priority sync downloads a subset of the data files (“priority files”) rather than all of the data files of the user, thereby minimizing the computing resources consumed to download the data files. The priority sync can select the priority files based on various criteria, including an access pattern of the data files, attributes of the data files, or attributes of the computing device. The priority sync can also download the data files not identified as priority files (“non-priority files”). While the priority files are downloaded in their original format, the non-priority files are downloaded in a downgraded format.

Abstract: Some examples include one or more processors that optimize the storage capacity of a computing device of a user. For instance, the one or more processors may receive, from the computing device, an indication of a request to download a data file. In addition, the one or more processors may determine a device classification associated with the computing device and a user access pattern associated with the data file. Further, the one or more processors may perform a downsampling of the data file based on the device classification and the user access pattern in response to the request, and may transmit the downsampled data file to the computing device.

Abstract: Technology is disclosed for enabling multiple computing devices used by a user to interoperate with one another. The technology enables the user to install a given application on a distributed operating system (“DCOS”) managing the multiple computing devices as a single entity. The user can then use the given application through any of the computing devices when executing the given application through the DCOS.

Abstract: Disclosed are systems, methods, and machine-readable storage mediums for selectively performing storage optimization processes in a computing device. In at least some embodiments, a method includes monitoring access patterns in association with data of one or more files accessed by a user of the computing device, determining a classification of the computing device, and utilizing the user access patterns and the device classification in determining execution of a storage optimization process based on a storage optimization policy. The storage optimization policy can specify downsampling, lossy compression, lossless compression, or a combination thereof to be applied to the files based on thresholds associated with the user access patterns and/or the device classification. Utilization of the storage optimization policy can enable the computing device to have an infinite local storage capacity.

Abstract: Some examples include caching data among multiple interconnected computing devices. As one example, a storage server may transmit a first data set to a first computing device, and may identifying a second computing device from a plurality of computing devices sharing a local area network with the first computing device. For instance, the second computing device may be selected from the plurality of computing devices based on battery levels of the plurality of computing devices. The storage server may identify a second data set as relevant to the first data set and may transmit the identified second data set to the second computing device. Additionally, the storage server may transmit an instruction to the first computing device indicating that the second computing device stores a data cache for the first computing device.

Abstract: Technology is disclosed for backing up and retrieving data in a distributed backup system (“the technology”). Files of a user can be stored across multiple computing devices (“the devices”) of the user, e.g., mobile devices, and/or at a server, e.g., a cloud storage server, in the distributed backup system. The user can define various policies for storing different files at different devices. For example, the user may define a policy for storing video files on a device that has a bigger display. The devices can be used as an edge cache of the distributed backup system in serving files to the user. Upon receiving a request for retrieving a file, the technology determines if any of the user devices that are in proximity to the requesting device has the file. If yes, the file is transmitted to the user from the proximate device instead of from the server.