Abstract: This disclosure describes a system for utilizing multiple image processing techniques to identify an item represented in an image. In some implementations, one or more image processing algorithms may be utilized to process a received image to generate item image information and compare the item image information with stored item image information to identify the item. When a similarity score identifying the similarity between the item image information and at least one of the stored item image information is returned, a determination may be made as to whether the similarity score is high enough to confidently identify the item. If it is determined that the similarity score is high enough to confidently identify the item, the other algorithms may be terminated and the determined identity of the item returned.

Abstract: Systems and methods for manage throttling limits in a distributed system are disclosed herein, according to some embodiments. A system includes a plurality of server nodes to perform a service. The system includes one or more processors a memory. The memory stores instructions that, when executed by the one or more processors, cause the one or more processors to perform operations. The operations include receiving a request for the service. The operations also include calculating whether accepting the request would exceed a service throttling limit for the plurality of server nodes for the service and whether accepting the request would exceed a node throttling limit for a server node of the plurality of server nodes. The operations also include accepting the request for processing at the server node responsive to calculating that the service throttling limit and the node throttling limit would not be exceeded.

Abstract: Technologies are provided for backing virtualization containers with layered storage volumes stored in remote storage devices. A virtualization container can present a virtual storage volume to a process running in the virtualization container and handle data access requests from the process using a layered storage volume based on one or more read-only snapshots stored in one or more remote storage devices. Changes can be recorded in a read-write layer in one or more of the remote storage devices and associated with the layered storage volume. A new read-only snapshot can be created based on the data in the read-write storage layer and associated with the one or more read-only snapshots on which the layered storage volume is based. A virtualization container can be associated with a new layered storage volume based on the created read-only snapshot and the one or more read-only snapshots on which it is based.

Abstract: Data within a database partition that is being migrated may be made available for access. Data within a database partition may be selected for migration. Writes to the selected data may be prevented while the data is being migrated. Writes to other data, or read access any of the partition, may be allowed during the migration of the selected data. Rollback events may be detected for a database partition that is being migrated, halting the further selection of data for migration and deleting the successfully migrated data at the copy of the database partition being created.

Abstract: Disclosed are various embodiments for customizing an item search and browsing experience based on a delivery speed preference. Multiple delivery speed options are generated for a client device, and a listing of the delivery speed options are sent to the client device. A user selection of a particular delivery speed option is then received from the client device. An item search criterion is received from the client device. A listing of items selected from an electronic item database is generated based at least in part on the item search criterion and item availability via the particular delivery speed option. Data encoding the listing of items is sent to the client device for rendering on a display.

Abstract: A system is disclosed that improves the ability for users to select links and/or other display elements via a touch screen, such as the touch screen of a smartphone, tablet, or other mobile device. The system achieves this effect by adjusting the sizes and/or shapes of the touch targets associated with particular display elements. For example, if a particular link on a web page is determined to be difficult to select via a touch screen (based on monitored user behaviors and/or based on an automated analysis of page content), the touch target associated with the link may be increased in size.

Abstract: Various systems and processes may be used for virtual disk importation. In particular implementations, systems and processes for virtual disk importation may include the ability to receive a request from a customer of a service provider network to import a virtual disk, which is accessible over a communication network, into the service provider network. The systems and processes may also include the ability to download a portion of a virtual disk file corresponding to the virtual disk, determine the type of the virtual disk by analyzing the downloaded portion, and determine the size of the virtual disk by analyzing the downloaded portion. The systems and processes may further include the ability to determine whether the virtual disk is supported by the service provider network based on the determined type of the virtual disk and download the virtual disk if the virtual disk is supported by the service provider network.

Abstract: A head-mounted wearable device (HMWD) may be fashioned that includes temples that pass near each ear of the user. A temple may incorporate a bone conduction speaker (BCS) to provide audio output to the user. During wear, a portion of the BCS is in physical contact with a head of the user to transfer the vibrations from the BCS to the user. To accommodate different head shapes and sizes, a head contact piece may be mounted to the BCS. The user may add, remove, or swap the head contact piece from an assortment of such pieces to find a profile that is comfortable to the wearer and improves performance of the BCS. The head contact piece may be magnetically affixed, allowing for easy changeout.

Abstract: Methods, systems, and computer-readable media for job scheduling based on job execution history are disclosed. A request is received to schedule a workload comprising a plurality of jobs. A resource allocation score for the workload is determined. The resource allocation score represents (at least in part) an estimated likelihood of successful execution of the workload. A first portion of the workload is scheduled for execution, and a remaining portion (if any) of the workload is delayed. A quantity of jobs in the first portion of the workload is determined based (at least in part) on the resource allocation score. Execution of the first portion of the workload is initiated.

Abstract: A rack computer system can provide data indicating electrical power consumption by separate sets of the mass storage devices, including separate individual mass storage devices, of the rack computer system. A power sensor can be electrically coupled to a power transmission line for each mass storage device. The power sensor can be coupled to the power transmission line externally to the mass storage device. The power sensor can be an internal power sensor of the mass storage device, where a mass storage device microcontroller transmits internally-generated data to an external power monitoring system. A microcontroller can transmit the data to a baseboard management controller via a side-band connection between the mass storage device and the controller. The data can be transmitted via an in-band connection between a baseboard management controller and an instance of firmware which accesses internally-generated data from mass storage device microcontrollers.

Abstract: A connectivity coordinator may receive a request for a dedicated physical connection between a provider network and a client network. In response, the connectivity coordinator may determine a capacity for each connection capacity group available to accept the dedicated physical connection. The connectivity coordinator may use the determined capacities for each available connection capacity group to determine bias response information for each connection capacity group based on the determined capacities and a capacity bias model. The connectivity coordinator may provide the determined bias response information to the client as part of a response to the connectivity request to influence the client to select to connect to a connection capacity group that has more physical ports available to accept the dedicated physical connection than other connection capacity groups that have less ports available to accept the dedicated physical connection.

Abstract: Systems and methods for providing object versioning in a storage system may support the logical deletion of stored objects. In response to a delete operation specifying both a user key and a version identifier, the storage system may permanently delete the specified version of an object having the specified key. In response to a delete operation specifying a user key, but not a version identifier, the storage system may create a delete marker object that does not contain object data, and may generate a new version identifier for the delete marker. The delete marker may be stored as the latest object version of the user key, and may be addressable in the storage system using a composite key comprising the user key and the new version identifier. Subsequent attempts to retrieve the user key without specifying a version identifier may return an error, although the object was not actually deleted.

Abstract: Apparatus and methods are disclosed herein for remote, direct memory access (RDMA) technology that enables direct memory access from one host computer memory to another host computer memory over a physical or virtual computer network according to a number of different RDMA protocols. In one example, a method includes receiving remote direct memory access (RDMA) packets via a network adapter, deriving a protocol index identifying an RDMA protocol used to encode data for an RDMA transaction associated with the RDMA packets, applying the protocol index to a generate RDMA commands from header information in at least one of the received RDMA packets, and performing an RDMA operation using the RDMA commands.

Abstract: Technologies are provided for secure sanitization of a storage device. A storage device can be configured to support an operational mode, into which the storage device is placed by default, and in which requests to cryptographically erase the storage device are rejected. The storage device can support a separate sanitization mode in which a request to cryptographically erase the storage device will be processed. Access to the sanitization mode can be restricted to trusted sources (such as a boot firmware of a computer connected to the storage device). The storage device can be configured to reject a command to place the storage device in the sanitization mode, unless the command is received during an initialization of the storage device. In at least some embodiments, the storage device can reject data access commands while it is in the sanitization mode.

Abstract: A block storage service can ensure volumes are placed in a same region as an attached virtual machine instance for performance and durability guarantees. A region can reference multiple things, but one example is that a volume is within a same spine as a virtual machine in order to meet performance guarantees. Each region can have a buffer of server computers held in reserve for volumes having a type where performance guarantees are required. If performance guarantees cannot be met, a rejection is transmitted to the customer. In another embodiment, the customer can provide a list in priority order of different volume types so that if a desired volume type cannot be placed, then alternative volume types can be used.

Abstract: A first set of execution platforms is deployed for a set of operations of a training phase of a machine learning model. Prior to the completion of the training phase, a triggering condition for deployment of a different set of execution platforms is detected. The different set of execution platforms is deployed for a subsequent set of training phase operations.

Abstract: A rack-mountable computer system directs separate portions of a cooling airflow from an inlet air plenum in the computer system interior through separate air passages to remove heat from separate portions of a set of heat-producing components. The air passages can preclude a portion of cooling airflow removing heat from a component from being preheated by another component. Plenums and air passages can be established through the arrangement of components in the interior. Components can be arranged in progressive offsets throughout the depth of the interior to vary plenum flow area throughout the depth, which can progressively impede and redirect airflow through the inlet plenum into the air passages. Arrangements can include an angled row, a staggered configuration, etc. The computer system can include a chassis which can translate, while maintaining operation of hot-pluggable electronic components coupled throughout the depth, to enable swapping of hot-pluggable electronic components throughout the interior.

Abstract: This disclosure describes a configuration of an unmanned aerial vehicle (“UAV”) in which the fuselage of the UAV is center mounted and at least some of the motors are configured to encompass at least a portion of the fuselage. In such a configuration, the stator and rotor of the motor extend around a perimeter of the fuselage, the propellers are coupled to an outer perimeter of the rotor, and the propellers extend radially outward away from the fuselage. Likewise, a closed wing may be coupled to the fuselage and positioned to encompass the radially extending propellers and at least a portion of the fuselage.

Abstract: A deployment subsystem provides updates to an application and/or software executed by server computer systems. The update is included in an update script. Execution of the update script by a server computer system causes the server computer system to execute a set of checks. After completing the set of checks the server computer system may execute the update to the application and/or software executed by server computer systems.

Abstract: Systems and methods are described for enabling cross-environment application of tracing information for code, such as code executed within an on-demand (or “serverless”) code execution system. Various optimizations exist that allow execution of code to proceed faster or more efficiently over time, by collecting tracing information regarding the execution and using that tracing information to guide compilation of the code. These optimizations are typically designed for long-lived environments. However, executions within an on-demand code execution system often occur in short-lived environments, reducing or eliminating any gains from these optimizations. To address this issue, tracing information can be maintained across multiple environments on the system, allowing subsequent executions to be optimized based on tracing information of prior executions in other environments.

Abstract: Described are systems, methods, and apparatus that enable constraint based communications between two or more devices. For example, a first user of a first device may submit a communication request to establish a communication session with a second user and provide a constraint for that communication session, such as a time-limit (e.g., limit the communication session to five minutes). In such an example, if the second user accepts the communication request with the constraint, a communication session is established and the system monitors the communication session to determine when a condition corresponding to the constraint has been satisfied. When the condition is satisfied, the communication session is terminated by the system.

Abstract: Methods, systems, and computer-readable media for dynamic and application-specific virtualized graphics processing are disclosed. Execution of an application is initiated on a virtual compute instance. The virtual compute instance is implemented using a server. One or more graphics processing unit (GPU) requirements associated with the execution of the application are determined. A physical GPU resource is selected from a pool of available physical GPU resources based at least in part on the one or more GPU requirements. A virtual GPU is attached to the virtual compute instance based at least in part on initiation of the execution of the application. The virtual GPU is implemented using the physical GPU resource selected from the pool and accessible to the server over a network.

Abstract: An advertisement request identifying a user may be received by a computing device of an electronic marketplace provider. User information for the user may be determined based on the advertising request. An advertisement featuring an item offered on an electronic marketplace may be selected. The advertisement may be provided for placement within content of a third-party network page provider. A first plurality of user input options configured to elicit a level of interest of the user with respect to the first advertisement may also be provided within the content of the third-party network page provider. User interaction information indicating the level of interest of the user with respect to the advertisement may be received. A second plurality of user input options may be determined based on the item of the advertisement and the first plurality of user input options. The second plurality of user input options may be provided to the content.

Abstract: A method and system for generating multiple profiles corresponding to different digital certificates. The profile includes intrinsic attributes and derived attributes associated with a digital certificate. The system enables a customer system to filter digital certificates based on a suitability of the various digital certificates for use with a given application to be executed by or on behalf of the customer system. The suitability may be determined based on a comparison of certificate requirements associated with a customer system's request and one or more of the intrinsic attributes and derived attributes.

Abstract: Systems and methods are provided herein for operating a conveyance system (e.g., a conveyor belt, tilt plane, robotic arm) of a mobile drive unit (MDU) independent of the MDU's navigation system. The MDU may be configured to obtain, convey, and deliver items within the workspace. Navigation information related to navigating the MDU within a workspace may be obtained. Operations of a navigational system of the MDU may be performed to move the MDU to various locations within the workspace in accordance with the navigation information. Conveyance information related to physically obtaining and/or delivering an item within the workspace may be obtained by the MDU (e.g., via conveyance-related fiducial markers and/or sensors of the MDU). At least one operation of a conveyance system of the MDU may be executed which may cause the item to be physically relocated based at least in part on the conveyance information.

Abstract: A number of approaches for transcoding media is described, which allow various users to control one or more aspects of transcoding (e.g. bit rate, resolution) when delivering media content to a client device from a network edge server or other location. The client device is provided with an interface that allows the user to increase or decrease the resolution, bit rate or other settings of the media content. Upon receiving a request from the user, the system transcodes the media content in accordance with the request and dynamically delivers the transcoded media content to the user's device. The system also enables content publishers and network operators to impose restrictions (e.g. minimum and maximum values for various settings) on user control.

Abstract: Certain embodiments herein relate to implementing collaborative bidding in online auctions. Example methods may include generating a bid request, receiving a first response comprising a first bid amount and a first content identifier, receiving a second response comprising a second bid amount and a second content identifier, applying a set of rules to the first response and the second response, and determining, using the set of rules, that the first bid amount and the second bid amount are combinable. Some example methods may include generating a combined bid amount using the first bid amount and the second bid amount, determining that the combined bid amount is a winning bid amount, determining a first allocated cost associated with the first response using the first bid amount, determining a second allocated cost associated with the second response using the second bid amount, and causing presentation of first content and second content.

Abstract: Session-specific information stored to a cookie or other secure token can be selected and/or caused to vary over time, such that older copies will become less useful over time. Such an approach reduces the ability of entities obtaining a copy of the cookie from performing unauthorized tasks on a session. A cookie received with a request can contain a timestamp and an operation count for a session that may need to fall within an acceptable range of the current values in order for the request to be processed. A cookie returned with a response can be set to the correct value or incremented from the previous value based on various factors. The allowable bands can decrease with age of the session, and various parameter values such as a badness factor for a session can be updated continually based on the events for the session.

Abstract: A network-based virtual computing resource provider may offer virtual compute instances that implement rolling resource credits for scheduling virtual computing resources. Work requests for a virtual compute instance may be received at a virtualization manager. A resource credit balance may be determined for the virtual compute instance. The resource credit balance may accumulate resource credits in rolling fashion, carrying over unused credits from previous time periods. Resource credits may then be applied when generating scheduling instructions to provide to a physical resource to perform the work requests, such as a physical CPU in order to increase the utilization of the resource according to the number of credits applied. Applied resource credits may then be deducted from the credit balance.

Abstract: A bus controller is configured to transmit a broadcast read request on at least one bus. The broadcast read request includes an address. A first logic module determines that the broadcast read request is targeting the first logic module. The first logic module reads a first value from a first register included in the first logic module. The first register is specified by the address included in the broadcast read request. The first value is transmitted onto the at least one bus. A second logic module determines that the broadcast read request is targeting the second logic module. The second logic module reads a second value from a second register included in the second logic module. The second register is specified by the address included in the broadcast read request. The second value is transmitted onto the at least one bus.

Abstract: This disclosure describes techniques for automating a system-level security review of a network-based service. The techniques may include generating and utilizing a machine-readable threat model to identify system-level security threats to the network-based service. The network-based service may be scanned upon being provisioned in a service-provider network, and the machine-readable threat model may be generated based on results of the scan. The machine-readable threat model may represent components of the network-based service, system-level security constraints configured to identify system-level security threats to the service, and mitigations to remedy violations to the system-level security constraints. The network-based service may be continuously, or periodically, scanned to identify changes in the network-based service.

Abstract: A centralized policy management may allow for one set of credentials to various applications and services offered by a computing resource service provider or other third-party servers. An entity responsible for the administration of a directory made available through a managed directory service may specify one or more policies for users and/or groups of users that utilize the directory. For example, the managed directory service may include a policy management subsystem that manages a set of policies for users and/or groups of users that controls a level of access to applications and services. Administrators can assign one or more policies to a user or a group of users and users can select one or more policies provided to the user by the administrator when attempting to access an application or service.

Abstract: Modifications to throughput capacity provisioned at a data store for servicing access requests to the data store may be performed according to cache performance metrics. A cache that services access requests to the data store may be monitored to collected and evaluate cache performance metrics. The cache performance metrics may be evaluated with respect to criteria for triggering different throughput modifications. In response to triggering a throughput modification, the throughput capacity for the data store may be modified according to the triggered throughput modification. In some embodiments, the criteria for detecting throughput modifications may be determined and modified based on cache performance metrics.

Abstract: A technology is described for redirecting a service request to a service proxy on a software development host. An example method may include receiving a service request directed to a service hosted in a service provider environment. In receiving the service request, a determination may be made that a service proxy simulates the service on a software development host, and that the service proxy may be used to execute development code on the software development host. In response, the service request may be redirected to the service proxy on the software development host for handling of the service request.

Abstract: An interactive system makes use of a concise user representation, for example, in the process of making predictions of a user's next action. In some examples, the concise user representation is computed from a larger amount of user data, which is processed using a transformation trained using a Generative Adversarial Network (GAN) approach.

Abstract: A tether compensated unmanned aerial vehicle (UAV) is described. In one embodiment, the UAV includes a winch with a tether to lower an item from the UAV for delivery, a tether compensation mechanism configured to contact the tether as it extends from the winch, and a flight controller to control a flight path of the UAV. The flight controller is also configured to direct the tether compensation mechanism to clamp the tether based on the flight path of the UAV. Further, based on movement identified in the tether using a sensor, a tether response controller can determine a complementary response and direct the tether compensation mechanism to brace the tether against the movement. Thus, the tether compensation mechanism can help stabilize sway or movement in the tether, which can help prevent the tether from undesirable swinging.

Abstract: A processing device receives a representation of an image, wherein the image has a first size and the representation has a second size that is smaller than the first size, the representation having been generated from the image by a first portion of a first trained machine learning model. The processing device processes the representation of the image using a second portion of the trained machine learning model to generate a reconstruction of the image and then outputs the reconstruction of the image.

Abstract: Embodiments of the disclosure are directed to a conveyor system that optimizes a gap between items. The gap is optimized to maximize throughput of the conveyor system with respect to bottlenecks in the conveyor system, such as label applicators that print and apply label to items that are being advanced through the conveyor system.

Abstract: Tiered storage may be implemented for processing data. Data processors may maintain some of a data set, including user data and metadata describing the user data, locally. The data set is also maintained a data store remote to the data processor. When processing requests are received, a determination is made as to whether the local portions of the data set can execute the processing request or one or more additional portions of the data set are needed from the remote data store. If additional portions of the data set are needed, then a request may be sent to the data store for the additional portions. Once received, the data processor may execute the processing request utilizing the additional portions. Portions of the data set maintained locally at the data processor may be selected and flushed from local storage to the remote data store.

Abstract: Techniques for making storage of data objects eventually durable using redundancy encoding are described herein. Data objects are stored in a first set of data storage devices with a first durability. After a predetermined length of time, the data objects are converted to data shards and distributed to a second set of data storage devices with a second durability that is distinct from the first durability.

Abstract: A determination is made that network access between a virtualized graphics device and a compute instance of a client is to be enabled. A source network address for graphics-related traffic of the compute instance is identified. From a range of source port numbers associated with the source network address, a particular source port number which is unused is found. Routing metadata is transmitted to one or more routing devices indicating that a key based at least in part on (a) the source network address and (b) the particular source port number is to be used to identify a route for network packets from the first application compute instance to a virtualized graphics device.

Abstract: Techniques and systems for storing and retrieving data storage devices of a data storage system are disclosed. In some embodiments, inventory holders are used to store data storage devices used by a data storage system. When data is to be transacted with the data storage devices, mobile drive units locate appropriate inventory holders and transport them to a device reading station, where an appropriate device retrieval unit transacts the data. In some embodiments, each inventory holder includes a heterogenous mix of data storage device types, the layout of which may be calculated according to the specific mix allocated to a given inventory holder. After the data has been transacted, the data storage devices are returned to the appropriate inventory holders, and the inventory holders are placed by the mobile drive units in locations where they may be accessed in response to further data transactions.

Abstract: Systems and methods are described to enable synchronized encoding of streaming audio or video content between multiple encoders, in a manner that provides for seamlessly interchangeable encodings. A first encoding begins encoding streaming content at a first point in time, and at a later point in time, a second encoder is configured to begin encoding the streaming content in a manner synchronized with the first encoder. The second encoder communicates via a synchronization protocol with the first encoder, and receives state information of the first encoder, such as a timecode of the streaming content at which the first encoder began encoding. The second encoder uses the received state information to determine a difference in encoding start times between the first and second encoder, and to adjust its timestamp values to match those of the first encoder. Thereafter, the first and second encoders can encode content in a synchronized manner.

Abstract: Routing data is copied from a primary data store to a local data store of a computing device. As the routing data is being copied by the computing device, changes to the data continue to occur. After copying all or a portion of the data, the computing device uses a value of a global sequence number (GSN) to determine if mutations have been made to the primary data. The routing data in the local data store is not consistent with the routing data stored in the primary data store when the value of the GSN is larger after the copying of the routing data as compared to the initial value of the GSN when the copying started. When changes have been made to the data, the computing device integrates the changes into the local data store before beginning routing operations.

Abstract: Dynamic application instance discovery and state management within a distributed system. A distributed system may implement application instances configured to perform one or more application functions within the distributed system, and discovery and failure detection daemon (DFDD) instances, each configured to store an indication of a respective operational state of each member of a respective group of the number of application instances. Each of the DFDD instances may repeatedly execute a gossip-based synchronization protocol with another one of the DFDD instances, where execution of the protocol between DFDD instances includes reconciling differences among membership of the respective groups of application instances. A new application instance may be configured to notify a particular DFDD instance of its availability to perform an application function.

Abstract: Some aspects of the present disclosure relate to generating and training a neural network by separating historical item interaction data into both inputs and outputs. This may be done, for example, based on date. For example, a neural network machine learning technique may be used to generate a prediction model using a set of inputs that includes both a number of items purchased by a number of users before a certain date as well as some or all attributes of those items, and a set of outputs that includes the items purchased after that date. The items purchased before that date and the associated attributes can be subjected to a time-decay function.

Abstract: A controller is configured to transmit a broadcast write request on at least one bus. The broadcast write request includes an address and a value. A first logic module determines that the broadcast write request is targeting the first logic module. The first logic module stores the value at a first addressed register specified by the register address. The second logic module determines that the broadcast write request is targeting the second logic module. The second logic module stores the value at a second addressed register specified by the register address. The first and second logic modules are connected to the at least one bus.

Abstract: An approach for interfacing a management control system with geographically dispersed electromechanical systems is described. Service messages are used to access services supported by a number of different types of electromechanical systems. The services of the systems are exposed through the service messages, and the systems are treated as service providers without the need to account for the different types, makes, models, and vendors of the systems when accessing them. Thus, similar to the way that a client device requests services from a server in a networked computing environment, the management control system can interact with the electromechanical systems to request services. The protocol of the service messages does not need to be changed as electromechanical systems are added, replaced, or updated, and new services and support for additional hardware can be added and updated as needed without changing the management control system.

Abstract: A system includes a provider network comprising resources of the provider network implemented on computing devices of the provider network and multiple networking devices. The networking devices are connected via physical network paths within the provider network. The system includes a traffic analyzer that receives at least two different sets of traffic information comprising different types of traffic measurements. The traffic analyzer determines traffic flowrates for the network paths of the provider network and/or a dedicated physical network path between the provider network and the customer network based on the at least two different sets of traffic information.

Abstract: A scalable encoding architecture is able to support multiple encoder versions without requiring new versions of the architecture to be generated and supported for each encoder version. An encoding profile template can be generated that includes the available settings for an encoder. A versioned schema can be generated that defines the settings for a particular encoder version. The schema can be persisted such that when a new channel or media stream is to be generated, the schema can be loaded and its values validated in order to cause the encoder to encode the media content using the appropriate values. An interface presented to the user to generate the template or channel can be dynamically generated such that only the applicable and/or relevant settings for a particular encoder version are displayed.