Technologies are described for determining between synchronous and asynchronous modes for computing service requests. Computing service requests are received by a computing service from clients. The computing service dynamically determines whether to use synchronous mode or asynchronous mode for processing the computing service requests. The computing service makes the dynamic determination of which mode to use (synchronous or asynchronous) based on various criteria, which can include synchronous/asynchronous mode recommendations generated by machine learning models and/or synchronous/asynchronous mode recommendations generated by static rules.

Techniques for packaging and deploying algorithms utilizing containers for flexible machine learning are described. In some embodiments, users can create or utilize simple containers adhering to a specification of a machine learning service in a provider network, where the containers include code for how a machine learning model is to be trained and/or executed. The machine learning service can automatically train a model and/or host a model using the containers. The containers can use a wide variety of algorithms and use a variety of types of languages, libraries, data types, etc. Users can thus implement machine learning training and/or hosting with extremely minimal knowledge of how the overall training and/or hosting is actually performed.

A method for application orchestration and management in a hybrid applications environment that comprises both VNF applications and CNF applications can include receiving an application descriptor that describes features of a new application to be deployed. A determination can be made, based at least in part on the application descriptor, whether the new application should be deployed as a VNF application or as a CNF application. The new application can be deployed as a VNF application when the application descriptor indicates that the new application should be deployed as a VNF application and when the available VNF infrastructure is sufficient to deploy the new VNF application. Conversely, the new application can be deployed as a CNF application when the application descriptor indicates that the new application should be deployed as a CNF application and when the available CNF infrastructure is sufficient to deploy the new CNF application.

Techniques described herein relate to a method for performing computational offloads for composed information handling systems. The method includes obtaining, by a system control processor associated with a composed information handling system, a computational offload request associated with a dataset from an application executing on an at least one compute resource set; in response to obtaining the computational offload request: identifying a dataset location associated with the dataset in the composed information handling system; identifying resources of the composed information handling system capable of performing the computational offload request; selecting a resource of the resources to perform the computational offload; and initiating performance of the computational offload request on the selected resource.

A device may receive a job request that requests performance of one or more operations by resources of a high-performance computing environment, and may process the job request, with a policy execution model trained with policy parameters, to identify policies to apply during execution of the job request. The device may process the job request, with a forecast object model trained with job data and profile data, to generate a forecast of resources and licenses required from the high-performance computing environment. The device may process the job request, other job requests, the one or more of the policies, and the forecast, with a heuristic model, to determine a schedule for the job request, and may process the schedule and current constraints on the resources and the licenses, with a linear programming model, to determine an optimized schedule for the job request.

Systems, computer program products, and methods are described herein for implementing data analytics in a mainframe environment. The present invention is configured to determine one or more data analytics resources associated with natural language processing algorithms; initiate one or more compiler protocols on the one or more data analytics resources to build one or more executable code for the one or more data analytics resources capable of being executed on a mainframe environment; establish a communication link with a job control language (JCL) subsystem associated with the mainframe environment; transmit the one or more executable code for the one or more data analytics resources to the JCL subsystem; generate one or more job control statements configured to be executable on the mainframe environment; generate a log of the one or more job control statements; and initiate an execution of the one or more job control statements on the mainframe environment.

According to an aspect, there is provided a computer-implemented method of operating a first node. The first node has an algorithm for evaluating input data from another node, with the input data having a plurality of different attributes. The method comprises receiving, from a second node, a proposal for the evaluation of a first set of input data by the algorithm; estimating the performance of the algorithm in evaluating the first set of input data based on the proposal; and outputting, to the second node, an indication of the estimated performance of the algorithm. A corresponding first node is also provided.

Embodiments are disclosed for emulation of graphics processing unit instructions. An example method executing an instrumented kernel using a logic circuit, the instrumented kernel including an emulation sequence; saving, in response to determination that the emulation sequence is to be executed, source data to a shared memory; setting an emulation request flag to indicate to processor circuitry separate from the logic circuit that offloaded execution of the emulation sequence is to be executed; monitoring the emulation request flag to determine whether the offloaded execution of the emulation sequence is complete; and accessing resulting data from the shared memory.

Code may be dynamically routed to computing resources for execution. Code may be received for execution on behalf of a client. Execution criteria for the code may be determined and computing resources that satisfy the execution criteria may be identified. The identified computing resources may then be procured for executing the code and then the code may be routed to the procured computing resources for execution. Permissions or authorization to execute the code may be shared to ensure that computing resources executing the code have the same permissions or authorization when executing the code.

A method for performing scheduling includes extracting information from at least one log file for an application. The method also includes determining an allocation of cloud resources for the application based on the information from the log file(s).