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: Migrating servers from client networks to virtual machines (VMs) on a provider network. A migration appliance is installed or booted on the client network, and a migration initiator is instantiated on the provider network. A VM and associated volumes are instantiated on the provider network. The initiator sends a request for a boot sector to the appliance; the appliance reads the blocks from a volume on the client network, converts the blocks to a format used by the VM, and sends the blocks to the initiator. The initiator boots the VM using the boot sector and the VM begins execution. The initiator then retrieves all data blocks for the VM from volumes on the client network via the appliance, stores the data to the volumes on the provider network, and fulfills requests from the VM from either local volumes or the remote volumes via the appliance.

Abstract: A first remote virtualized graphics device is instantiated in response to a determination that processing of graphics operations is to be enabled in a first availability mode on behalf of a compute instance. A configuration operation is performed at a routing device to enable packets from the first remote virtualized graphics device to be transmitted to a graphics result destination. In response to an indication of unavailability, the routing device is configured to enable packets from a second remote virtualized graphics device to be directed to the graphics result destination.

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: 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: Methods, systems, and computer-readable media for interaction monitoring for virtualized graphics processing are disclosed. Execution of an application is initiated on a virtual compute instance that is implemented using CPU and memory resources of a server. Instruction calls are produced by the execution of the application and sent from the server to a graphics server over a network. The graphics server comprises a physical GPU, and a virtual GPU is implemented using the physical GPU and attached to the virtual compute instance. GPU output is generated at the graphics server based at least in part on execution of the instruction calls using the virtual GPU. A log of interactions between the application and the virtual GPU is stored. The interactions comprise the instruction calls sent to the graphics server and responses to the instruction calls sent to the virtual compute instance.

Abstract: Methods, systems, and computer-readable media for interaction monitoring for virtualized graphics processing are disclosed. Execution of an application is initiated on a virtual compute instance that is implemented using CPU and memory resources of a server. Instruction calls are produced by the execution of the application and sent from the server to a graphics server over a network. The graphics server comprises a physical GPU, and a virtual GPU is implemented using the physical GPU and attached to the virtual compute instance. GPU output is generated at the graphics server based at least in part on execution of the instruction calls using the virtual GPU. A log of interactions between the application and the virtual GPU is stored. The interactions comprise the instruction calls sent to the graphics server and responses to the instruction calls sent to the virtual compute instance.

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: 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: Migrating servers from client networks to virtual machines (VMs) on a provider network. A migration appliance is installed or booted on the client network, and a migration initiator is instantiated on the provider network. A VM and associated volumes are instantiated on the provider network. The initiator sends a request for a boot sector to the appliance; the appliance reads the blocks from a volume on the client network, converts the blocks to a format used by the VM, and sends the blocks to the initiator. The initiator boots the VM using the boot sector and the VM begins execution. The initiator then retrieves all data blocks for the VM from volumes on the client network via the appliance, stores the data to the volumes on the provider network, and fulfills requests from the VM from either local volumes or the remote volumes via the appliance.

Abstract: Methods, systems, and computer-readable media for dynamic interface synchronization for virtualized graphics processing are disclosed. A GPU interface synchronization request is sent from a compute instance to a graphics processing unit (GPU) server via a network. The GPU server comprises a virtual GPU attached to the compute instance and implemented using at least one physical GPU. Based at least in part on the GPU interface synchronization request, a shared version of a GPU interface is determined for use with the compute instance and the GPU server. Program code of the shared version of the GPU interface is installed on the compute instance and on the GPU server. Using the shared version of the GPU interface, the compute instance sends instructions to the virtual GPU over the network, and the virtual GPU generates GPU output associated with the instructions.

Abstract: Migrating servers from client networks to virtual machines (VMs) on a provider network. A migration appliance is installed or booted on the client network, and a migration initiator is instantiated on the provider network. A VM and associated volumes are instantiated on the provider network. The initiator sends a request for a boot sector to the appliance; the appliance reads the blocks from a volume on the client network, converts the blocks to a format used by the VM, and sends the blocks to the initiator. The initiator boots the VM using the boot sector and the VM begins execution. The initiator then retrieves all data blocks for the VM from volumes on the client network via the appliance, stores the data to the volumes on the provider network, and fulfills requests from the VM from either local volumes or the remote volumes via the appliance.

Abstract: Methods and apparatus for updating virtual machines (VMs) on a provider network according to modifications made to a server in a client network. A version of the server may be currently instantiated and executing as one or more VM instances on the provider network. Agent(s) installed on the server in the client network intercept write requests to volume(s) attached to the server, and send blocks that include updates to the server volume(s) to a service on the provider network. The service stores the blocks to incremental snapshots, and generates timestamped machine images (MIs) of the server from the snapshots. A VM service updates the VM instances on the provider network according to the MIs. Thus, the VM instances can be kept up to date with changes to the server without having to upload the entire volume(s) to the provider network to perform each update.

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: A service provider system may implement ECC-like features when executing computations on GPUs that do not include sufficient error detection and recovery for computations that are sensitive to bit errors. During execution of critical computations on behalf of customers, the system may automatically instrument program instructions received from the customers to cause each computation to be executed using multiple sets of hardware resources (e.g., different host machines, processor cores, or internal hardware resources). The service may provide APIs with which customers may instrument their code for execution using redundant resource instances, or specify parameters for applying the ECC-like features. The service or customer may instrument code to perform (or cause the system to perform) checkpointing operations at particular points in the code, and to compare intermediate results produced by different hardware resources.

Abstract: A provider network may implement resource credit pools to replenish resource credit balances for virtual compute instances. A resource credit pool may be maintained that makes resource credits available to virtual compute instances authorized to obtain resource credits from the resource credit pool. Resource credits from the resource credit pool may be applicable to increase utilization of physical computer resource for a virtual compute instance. In response to a resource credit request for an authorized virtual compute instance, a number of resource credits to add to an individual resource credit balance for the authorized virtual compute instance may be determined. A response may be sent indicating the number of resource credits to add to the individual resource credit balance and the resource credit pool may be updated to remove the number of resource credits from the resource credit pool.

Abstract: A virtualization host may implement variable timeslices for processing latency dependent workloads. Multiple virtual compute instances on a virtualization host may utilize virtual central processing units (vCPUs) to obtain physical processing resources, such as one or more central processing units (CPUs). A vCPU currently utilizing a CPU to performing processing work according to a scheduled timeslice may be preempted by a latency dependent vCPU before completion of the scheduled timeslice. The latency-dependent vCPU may complete processing work, and utilization of the CPU may be returned to the vCPU. A preemption compensation may be determined for the scheduled timeslice to increase the scheduled timeslice for the vCPU such that utilization for the vCPU is performed according to the increased scheduled timeslice.

Abstract: A provider network may implement resource credit pools to replenish resource credit balances for virtual compute instances. A resource credit pool may be maintained that makes resource credits available to virtual compute instances authorized to obtain resource credits from the resource credit pool. Resource credits from the resource credit pool may be applicable to increase utilization of physical computer resource for a virtual compute instance. In response to a resource credit request for an authorized virtual compute instance, a number of resource credits to add to an individual resource credit balance for the authorized virtual compute instance may be determined. A response may be sent indicating the number of resource credits to add to the individual resource credit balance and the resource credit pool may be updated to remove the number of resource credits from the resource credit pool.

Abstract: A virtualization host may implement variable timeslices for processing latency dependent workloads. Multiple virtual compute instances on a virtualization host may utilize virtual central processing units (vCPUs) to obtain physical processing resources, such as one or more central processing units (CPUs). A vCPU currently utilizing a CPU to performing processing work according to a scheduled timeslice may be preempted by a latency dependent vCPU before completion of the scheduled timeslice. The latency-dependent vCPU may complete processing work, and utilization of the CPU may be returned to the vCPU. A preemption compensation may be determined for the scheduled timeslice to increase the scheduled timeslice for the vCPU such that utilization for the vCPU is performed according to the increased scheduled timeslice.

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.