Abstract: A method for managing virtual machines includes selecting a virtual machine (VM) on a recoverable deleted VMs List, and initiating recovery of the VM on the production host.

Abstract: Techniques are described for optimizing the allocation of computing resources provided by a service provider network—for example, compute resources such as virtual machine (VM) instances, containers, standalone servers, and possibly other types of computing resources—among computing workloads associated with a user or group of users of the service provider network. A service provider network provides various tools and interfaces to help businesses and other organizations optimize the utilization of computing resource pools obtained by the organizations from the service provider network, including the ability to efficiently schedule use of the resources among workloads having varying resource demands, usage patterns, relative priorities, execution deadlines, or combinations thereof. A service provider network further provides various graphical user interfaces (GUIs) to help users visualize and manage the historical and scheduled uses of computing resources by users' workloads according to user preferences.

Abstract: A method for managing AI components installed in containers is provided. The container-based component management method creates a container, installs at least one selected from a plurality of components in the container, and manages the components installed in the container. Accordingly, the execution priorities of the AI components installed in the containers can be managed and operated, such that degradation of system performance and frequent error occurrence can be prevented.

Abstract: On a device, application usage information is monitored. From an analysis of the application usage information, an application usage pattern information is constructed. A pattern in the application usage pattern information identifies a second application in a second set of applications, wherein the second application is concurrently active with a first application in the first set of applications during an activity on the device. From a current activity on the device, the current activity having a degree of similarity above a threshold degree of similarity to the activity of the pattern in the application usage pattern information is detected. For user interaction responsive to the detecting, at least two applications selected based on the pattern are activated.

Abstract: In a computer-implemented method for generating a temporal topology graph of a computing environment, managed component relationship data is received for a plurality of managed components of a computing environment, the managed component relationship data including parent/child information for a managed component of the plurality of managed components at a moment in time. The managed component relationship data is transformed into graphical data of a temporal topology graph of the computing environment, wherein the temporal topology graph includes the moment in time for each instance of the managed component relationship data within the temporal topology graph. The temporal topology graph including the managed component relationship data is generated based at least in part on the graphical data. The temporal topology graph for the computing environment is maintained at a service provider.

Abstract: A driver loading method and a server, where when receiving a service request, the server determines a first global index and a first global virtual function (VF) identifier corresponding to a first function description of a designated function included in the service request, determines a virtual machine (VM) corresponding to the service request, associates the first global VF identifier with the VM, allocates a first local index on the VM to the designated function, creates a correspondence between the first local index and the first function description, and sends the correspondence to the VM. The VM loads, according to the correspondence, a driver of the designated function for a first VF corresponding to the first global VF identifier. According to the foregoing method, different drivers can be loaded for VFs that have different functions and that are virtualized by a Peripheral Component Interconnect Express (PCIe) device.

Abstract: Systems and methods of scheduling for Network Function Virtualization (NFV) on processing resources include receiving multiple flows from different customers and different applications; classifying the multiple flows to provide classified flows; distributing the classified flows to a plurality of queues; and providing each of the classified flows in the plurality of queues to the processing resources, wherein each individual classified flow is distributed to a same processing resource thereby maintaining flow order of the individual classified flow.

Abstract: A technique is introduced for intercepting memory calls from a user-space application and applying an allocation policy to determine whether such calls are handled using volatile memory such as dynamic random-access memory (DRAM) or persistent memory (PMEM). In an example embodiment, memory calls from an application are intercepted by a memory allocation capture library. Such calls may be to a memory function such as malloc( ) and may be configured to cause a portion of DRAM to be allocated to the application to process a task. The memory allocation capture library then determines whether the intercepted call satisfies capture criteria associated with an allocation policy. If the intercepted call does satisfy the capture criteria, the call is processed to cause a portion of PMEM to be allocated to the application instead of DRAM.

Abstract: A method for executing, in a distributed computing infrastructure (comprising a broker node and work nodes), a job workload having a requirement related to a resource. Test deployment comprises: executing a test workload observing the requirement and varying the resource; for each test workload execution, receiving a metric quantifying a response of the test node to the status of the resource. The broker node: performs the test deployment with the test workload being a measuring workload having the requirement, the test node being a first and second candidate node, and the metric respectively being a first and second metric; determines a cost impact function of the metric; respectively determines a minimum of a first and second total cost function determined from the first and second metric and the cost impact function; and causes an executing node selected from the candidate nodes based on the minima to execute the job workload.

Abstract: Systems and methods are disclosed for batch job execution. In some embodiments, a method may include: executing a plurality of service instances, each service instance being configured to execute jobs stored in a database; and by a first service instance of the plurality of service instances, identifying a job stored in the database, determining that no service instance of the plurality of service instances other than the first service instance has initiated execution of the job, generating an execution timestamp having a granularity of one second or longer, recording the generated execution timestamp in the database such that the execution timestamp is associated with the job, and executing the job. The service instances other than the first service instance may be configured to avoid execution of the job upon determining that a respective system time matches the execution timestamp to the granularity of the execution timestamp.

Abstract: Method and system embodying the method for scheduling arbitration among a plurality of service requestors encompassing: designating among the plurality of service requestors all the service requestors that have an active request; determining whether at least one of the designated service requestors has an un-served status indicator which is set; and when the determining is positive then: selecting one of the at least one designated service requestors in accordance with a pre-determined policy; and clearing the un-served status indicator for the selected service requestor, is disclosed.

Abstract: A method, an information handling system (IHS), and a host group generating system for collecting information about host groups in a cluster of IHSs using a management controller. The method includes retrieving, from a controller memory, first target physical hosts that are part of a virtual machine manager (VMM) host group, but not on a VMM host list and retrieving second target physical hosts that are part of the VMM host list, but not part of the VMM host group. The method further includes transmitting, to each of the first target physical hosts, a request to detach from the VMM host group. The method further includes transmitting, to each of the second target physical hosts, a request to attach to the VMM host group.

Abstract: A system and method for dynamic halt-polling control includes a memory, one or more physical processors in communication with the memory, and a hypervisor executing on the one or more physical processors. In an example, the hypervisor enables halt-polling and disables monitoring of a guest in a virtual machine. Then, the hypervisor exposes the monitoring to the guest. In an example, the hypervisor detects a monitoring instruction executed on a virtual central processing unit (VCPU) of the virtual machine. The monitoring instruction is executed by the guest, which causes an exit to the hypervisor while the monitoring is disabled. Responsive to detecting the monitoring instruction that causes the exit to the hypervisor, the hypervisor disables the halt-polling and enables the monitoring of the guest.

Abstract: A method and system for improving virtual machine allocation and migration is provided. The method includes initiating a migration process for migrating database files of a virtual machine from a first hardware device to a second hardware device. A checkpoint and restart command is transmitted to a first hypervisor of the first hardware device and a request for a cryptographic key from a memory encryption engine is received. The cryptographic key is transmitted to a first enclave and the first enclave is encrypted resulting in an encrypted enclave. A resulting a data file comprising the database files is generated and the encrypted enclave is disconnected from the first hardware device. The encrypted enclave is destroyed and checkpoint and restart code is executed for restarting the first hardware device.

Abstract: A resource assignment method, and a recording medium and a distributed processing device applying the same are provided. The resource assignment method includes: when information regarding a plurality of tasks is received from a plurality of first nodes, calculating a size of a resource necessary for each of the received plurality of tasks; and when information regarding an available resource is received from a second node, assigning one of the plurality of tasks to the available resource of the second node, based on the calculated size of the resource necessary for each task.

Abstract: A method of assigning I/O requests to CPU cores of a data storage system includes, in a first operating mode, assigning I/O requests to CPU cores based on port affinity while maintaining a current I/O completion count, and regularly performing a first test-and-switch operation that includes (i) for a sample interval, temporarily assigning the I/O requests to the CPU cores based on core availability while obtaining a sample I/O completion count, (ii) comparing the first sample I/O completion count to the current I/O completion count, and (iii) based on the sample I/O completion count being greater than the current I/O completion count, switching to a second operating mode. In the second operating mode, I/O requests are assigned to the CPU cores based on core availability, and similar operations are performed for periodically testing whether to switch to the first operating mode.

Abstract: A service scheduling method, including: obtaining scheduling information of multiple services deployed on a network device; generating scheduling logic according to the scheduling information, invoking, according to the generated scheduling logic, each processing module to process a packet received by the network device, and invoking, according to the scheduling point information of each service, a corresponding service at a scheduling point of each service. Accordingly, the embodiments of the present invention also provide a service scheduling apparatus and a network device.

Abstract: A microprocessor system (1) includes a host processor (2), a graphics processing unit (GPU) (3) that includes a number of processing cores (4), and an exception handler. When a thread that is executing on a processing core (4) encounters an exception in its instruction sequence, the thread is redirected to the exception handler. However, the exception event is also communicated to a task manager (5) of the GPU 3. The task manager (5) then broadcasts a cause exception message to each processing core (4). Each processing core then identifies the threads that it is currently executing that the cause exception message relates to, and redirects those threads to the exception handler. In this way, an exception caused by a single thread is broadcast to all threads within a task.

Abstract: A method and apparatus for starting a virtual machine. A specific implementation of the method comprises: acquiring, by a physical machine, a mirror image file required for starting a to-be-started target virtual machine from a distributed block storage system, in response to an entered instruction to start the target virtual machine; and starting the target virtual machine by using the mirror image file. The mirror image file required for starting the virtual machine is stored in the cloud-based distributed block storage system, and a virtual disk is mapped to the physical machine. When the physical machine needs to start the virtual machine, the mirror image file required for starting the virtual machine is acquired from the cloud-based distributed block storage system by reading the virtual disk.

Abstract: The present disclosure provides systems and methods for domain decomposition in parallel numerical computer modelling that remove the need for large-scale synchronization, decouple the regions within a simulation, and reduces wait times, increasing linear scalability and thus suitability for large-scale cloud deployment. Presently disclosed systems and methods include asynchronous message passing which decouples the computational units within the system and introduces an elasticity that allows wait time to be largely eliminated. Applications include domains in which computations are not easily parallelized and/or large-scale numerical simulations used with a large cluster or cloud computing facility.