Abstract: Methods, apparatus, systems, and articles of manufacture for allocating a workload to an accelerator using machine learning are disclosed. An example apparatus includes a workload attribute determiner to identify a first attribute of a first workload and a second attribute of a second workload. An accelerator selection processor causes at least a portion of the first workload to be executed by at least two accelerators, accesses respective performance metrics corresponding to execution of the first workload by the at least two accelerators, and selects a first accelerator of the at least two accelerators based on the performance metrics. A neural network trainer trains a machine learning model based on an association between the first accelerator and the first attribute of the first workload. A neural network processor processes, using the machine learning model, the second attribute to select one of the at least two accelerators to execute the second workload.

Abstract: Optimizing a performance of a software function withing a content distribution network, such as a software-implemented virtual cable modem termination system (CMTS) network, a virtualized Radio Access Network (vRAN), a Passive Optical Network (PON), or a Wi-Fi network. The performance may be optimized by dynamically changing a deployment location within the content distribution network for the software function from an original location to an updated location using an instance management platform. The deployment location may be dynamically changing in response to a variety of trigger conditions or concerns, such as but not limited to a difference in compute resources, responding to latency needs or tolerances, and a desired cohabitation of data or other software.

Abstract: Systems and methods for providing a guaranteed batch pool are described, including receiving a job request for execution on the pool of resources; determining an amount of time to be utilized for executing the job request based on available resources from the pool of resources and historical resource usage of the pool of resources; determining a resource allocation from the pool of resources, wherein the resource allocation spreads the job request over the amount of time; determining that the job request is capable of being executed for the amount of time; and executing the job request over the amount of time, according to the resource allocation.

Abstract: A hierarchical resource management system for a building includes one or more processors. The processors implement a plurality of agents that each monitor sensed values, and generate operating scenarios based on the sensed values for corresponding resources. The processors also implement a coordinator that filters the operating scenarios to remove the operating scenarios that violate internal laws of the agents to form an aggregate validated set of operating scenarios. The processors further implement a supervisor that, responsive to receipt of target conditions for the zones and the aggregate validated set of operating scenarios from the coordinator, selects a combination of the operating scenarios from the aggregate validated set of operating scenarios that achieves target conditions and minimizes overall energy consumption by the resources such that some of the operating scenarios of the combination do not minimize energy consumption of the resources corresponding to the some of the operating scenarios.

Abstract: A software switch and a method performed by the software switch are disclosed. The software switch receives, from a node deploying a virtual machine, a request for a virtual port to be polled by the virtual machine. The request includes a Central Processing Unit “CPU” identity identifying a CPU on which the virtual machine executes. The request includes an indication of a clock frequency at which the CPU is set to operate. The software switch determines a number of packets in a queue associated with the virtual port. The software switch adjusts the clock frequency of the CPU based on the number of packets in the queue. A corresponding computer program and a computer program carrier are also disclosed.

Abstract: Disclosed is a system and method for cloud workload provisioning. In one implementation, the present invention provides a system enabling an automated guidance to the user for the workload to be provisioned. The present invention matches the user's workload profile based on a wide variety of historical data set and makes easy for users to choose the cloud provisioning for various kinds of workloads. The system can automatically readjust a workload profile for cloud provisioning. The system can provide a manual selection option for cloud provisioning. In one embodiment, the present invention provides a system and method that derives a workload provision scaling factor mechanism using historic data set. Furthermore, the system and method can automatically or manually readjust the provision scaling factor based on a workload profile for cloud provisioning.

Abstract: Systems and methods for redirecting one or more printers from a virtual computing device to a physical computing device. The method includes initiating a printing application on a virtual computing device and exporting at least one printer configuration. The at least one printer configuration corresponds to one or more printers communicatively coupled to a physical computing device. The method also includes storing the at least one printer configuration to a data file on the virtual computing device and copying the data file from the virtual computing device to the physical computing device. The method further includes invoking the data file on the physical computing device such that the at least one printer configuration is imported on the physical computing device. The method also includes generating for display on the virtual computing device at least one indication of the one or more printers communicatively coupled to the physical computing device.

Abstract: A computerized-method using a cloud-based computing environment for improving client service, in a contact center is provided herein. The computerized-method includes: retrieving a context of a query and a time-limit from a CTI event and attempting to retrieve data to evaluate average resolution time for the received context. When the data is found, comparing the evaluated average resolution time with the received time-limit and when the received time-limit is below the evaluated average resolution time, sending a delay notice and providing the client a menu of options for querying through other channels. When the data is not found, or when the received time-limit is above the evaluated average resolution time, presenting on an agent dashboard, the time-limit of the client and accordingly updating parameters in the agent dashboard during the inbound call, thus, improving client service, by considering the time-limit of the client before the agent addresses a query.

Abstract: A resource management system includes a resource management device that manages a usage of a resource, a reservation management device that manages a reservation for the resource, and a communication relay device that communicates with the resource management device. The resource management device receives reservation information transmitted from the reservation management device. The reservation information is information on the reservation for the resource. The resource management device transmits, to the communication relay device, a request for identification information identifying the communication relay device. The resource management device receives the identification information, which is transmitted from the communication relay device in response to the request. The identification information is used to execute an event associated with the reservation information.

Abstract: A combinatorial solution determination system includes a candidate solution generation unit that generates candidate combinatorial solutions, a simulation unit, an evaluation index calculation unit, a solution determination unit, and a necessary-series-data extraction unit. The simulation unit calculates simulation data using information related to the candidate combinatorial solutions, and series data to evaluate a combinatorial solution. The evaluation index calculation unit calculates evaluation indices based on the simulation data. The solution determination unit determines a combinatorial solution having high evaluation from the candidate combinatorial solutions based on the evaluation indices each calculated by the evaluation index calculation unit from a corresponding one of the candidate combinatorial solutions.

Abstract: Disclosed herein are methods and systems for providing data recovery recommendations. In an exemplary aspect, a method may comprise identifying a plurality of storage devices. For each respective device of the plurality of storage devices, the method may comprise extracting a respective input parameter indicative of a technical attribute of the respective device, inputting the respective input parameter into a machine learning algorithm configured to output both a first likelihood of the respective device needing a data recovery and a second likelihood that the data recovery will fail, and determining a respective priority level of the respective device based on the first likelihood and the second likelihood. The method may comprise normalizing each respective priority level, and recommending a device of the plurality of storage devices for a test data recovery procedure based on each normalized priority level.

Abstract: Deep reinforcement learning techniques are provided for resource allocation in a shared computing environment.

Abstract: A request is received to schedule a new software process. Description data associated with the new software process is retrieved. A workload resource prediction is requested and received for the new software process. A landscape directory is analyzed to determine a computing host in a managed landscape on which to load the new software process. The new software process is executed on the computing host.

Abstract: A processor may run a background process to identify a first task being initiated by a first user on a device, where the first task is associated with a first application. The processor may identify the first user of the device. The processor may analyze one or more interactions of the first user associated with the first application on the device. The processor may allocate, based at least in part on identification of the first user, identification of the first task, or analysis of the one or more interactions of the first user, computing resources to one or more hardware components on the device.

Abstract: A method including obtaining resource overheads according to feature information of a received query request; according to the resource overheads and a compute node resource, dynamically adjusting a compute node in a resource pool; and querying, by using the compute node, data corresponding to the query request. A compute node in a resource pool may be dynamically adjusted, so that the compute node in the resource pool may process all the received query requests, and therefore, the processing efficiency and a resource utilization rate of the compute node are more effectively improved, such that the compute node may more efficiently perform parallel processing on the multiple query requests, and the utilization rates of a CPU resource, a memory resource and a network bandwidth resource are increased, thus achieving better effect from the perspectives of overall computing resource and user query load and improving the usage experience of a user.

Abstract: This application describes a virtual machine deployment method and an operation and maintenance management (OMM) virtual machine. The method includes: obtaining, by an OMM virtual machine, a quantity and a specification of service virtual machines created in virtual network function application software to which the OMM virtual machine belongs; and determining, by the OMM virtual machine based on load that needs to be carried by the application software and the quantity and the specification of the service virtual machines, a module to be configured for each service virtual machine. The described implementations avoid or reduce waste of virtual machine resources.

Abstract: Method and system are provided for writing output from map-reduce jobs to an archive file. The method may include providing an archive manager and exposing an interface to be called from map-reduce jobs to output to an archive file in a map-reduce distributed file system. The method may also include using a buffering database as a temporary cache to buffer updates to the archive file. Handling by the archive manager calls from map-reduce jobs may allow: reading directly from an archive file or from a job index at the buffering database; writing to a job index at the buffering database used as a temporary cache to buffer updates; and serializing updates from the buffering database to the archive file.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for assigning operations of a computational graph to a plurality of computing devices are disclosed. Data characterizing a computational graph is obtained. Context information for a computational environment in which to perform the operations of the computational graph is received. A model input is generated, which includes at least the context information and the data characterizing the computational graph. The model input is processed using the machine learning model to generate an output defining placement assignments of the operations of the computational graph to the plurality of computing devices. The operations of the computational graph are assigned to the plurality of computing device according to the defined placement assignments.

Abstract: Aspects of the subject disclosure may include, for example, transmitting a first identifier associated with a processing system to a communication device, transmitting first data to the communication device, obtaining, from the communication device and based on the transmitting of the first identifier and the first data, a first directive, wherein the first directive directs the processing system to utilize a first radio access technology for a first communication session, and utilizing, based on the first directive, the first radio access technology for the first communication session. Other embodiments are disclosed.

Abstract: A service provider network implements seamless scaling via proxy replay of session state. Upon a trigger, such as a determination to scale a server, a scaled server may be spun up and an identifier of the scaled server provided to a first (existing) server. The first server sends the identification of the second server, and session state information for each of the connections between the first server and the request router, to the request router. For each of the connections, the request router establishes a new connection between the request router and the second (scaled) server, and replays the session state information for the connection to the second server. The request router then routes traffic between each existing client connection (e.g., the same existing client connection which carried traffic delivered to the first server) and the corresponding new connection to the second server.

Abstract: Various techniques are described for classifying synthetic data tasks and orchestrating a resource allocation between groups of eligible resources for processing the synthetic data tasks. Received synthetic data tasks can be classified by identifying a task category and a corresponding group of eligible resources (e.g., processors) for processing synthetic data tasks in the task category. For example, synthetic data tasks can include generation of source assets, ingestion of source assets, identification of variation parameters, variation of variation parameters, and creation of synthetic data. Certain categories of synthetic data tasks can be classified for processing with a particular group of eligible resources. For example, tasks to ingest synthetic data assets can be classified for processing on a CPU only, while a task to create synthetic data assets can be classified for processing on a GPU only. The synthetic data tasks can be queued and routed for processing by an eligible resource.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, are described for scheduling tasks of ML workloads. A system receives requests to perform the workloads and determines, based on the requests, resource requirements to perform the workloads. The system includes multiple hosts and each host includes multiple accelerators. The system determines a quantity of hosts assigned to execute tasks of the workload based on the resource requirement and the accelerators for each host. For each host in the quantity of hosts, the system generates a task specification based on a memory access topology of the host. The specification specifies the task to be executed at the host using resources of the host that include the multiple accelerators. The system provides the task specifications to the hosts and performs the workloads when each host executes assigned tasks specified in the task specifications for the host.

Abstract: Techniques are described for a data recovery validation test. In examples, a processor receives a command to be included in the validation test that is configured to validate performance of an activity by a server prior to a failure to perform the activity by the server. The processor stores the validation test including the command on a memory device, and prior to the failure of the activity by the server, executes the validation test including the command responsive to an input. The processor receives results of the validation test corresponding to the command and indicating whether the server performed the activity in accordance with a standard for the activity during the validation test. The processor provides the results of the validation test in a user interface.

Abstract: The present disclosure relates to dynamically scheduling resource requests in a distributed system based on usage quotas. One example method includes identifying usage information for a distributed system including atoms, each atom representing a distinct item used by users of the distributed system; determining that a usage quota associated with the distributed system has been exceeded based on the usage information, the usage quota representing an upper limit for a particular type of usage of the distributed system; receiving a first request for a particular atom requiring invocation of the particular type of usage represented by the usage quota; determining that a second request for a different type of usage of the particular atom is waiting to be processed; and processing the second request for the particular atom before processing the first request.

Abstract: This document relates to orchestration and scheduling of services. One example method involves obtaining dependency information for an application. The dependency information can represent data dependencies between individual services of the application. The example method can also involve identifying runtime characteristics of the individual services and performing automated orchestration of the individual services into one or more application processes based at least on the dependency information and the runtime characteristics.

Abstract: Systems, apparatuses, and methods for efficiently sharing registers among threads are disclosed. A system includes at least a processor, control logic, and a register file with a plurality of registers. The processor assigns a base set of registers to each thread of a plurality of threads executing on the processor. When a given thread needs more than the base set of registers to execute a given phase of program code, the given thread executes an acquire instruction to acquire exclusive access to an extended set of registers from a shared resource pool. When the given thread no longer needs additional registers, the given thread executes a release instruction to release the extended set of registers back into the shared register pool for other threads to use. In one implementation, the compiler inserts acquire and release instructions into the program code based on a register liveness analysis performed during compilation.

Abstract: In accordance with an embodiment, described herein is a system and method for use in a distributed computing environment, for automatically scaling a cluster based on metrics being monitored. A cluster that comprises a plurality of nodes or brokers and supports one or more colocated partitions across the nodes, can be associated with an exporter process and alert manager that monitors metrics associated with the cluster. Various metrics can be associated with user-configured alerts that trigger or otherwise indicate the cluster should be scaled. When a particular alert is raised, a callback handler associated with the cluster, for example an operator, can automatically bring up one or more new nodes, that are added to the cluster, and then reassign a selection of existing colocated partitions to the new nodes/brokers, such that computational load can be distributed within the newly-scaled cluster environment.

Abstract: The present invention is a system and method of continuous sentiment tracking and the determination of optimized agent actions through the training of sentiment models and applying the sentiment models to new incoming interactions. The system receives conversations comprising incoming interactions and agent actions and determines customer sentiment on a micro-interaction level for each incoming interaction. Based on interaction types, the system correlates the determined sentiment with the agent action received prior to the sentiment determination to create and train sentiment models. Sentiment models include agent action recommendations for a desired sentiment outcome. Once trained, the sentiment models can be applied to new incoming interactions to provide CSRs with actions that will yield a desired sentiment outcome.

Abstract: Improving load distribution and consistency is provided. A device intermediary to clients and servers can maintain bit values indicative of server availability stored in indices arranged in various levels. A lowest level comprises indices corresponding to a list of servers repeated multiple times. Each index in a higher level maps to a set of indices in a lower level. The device can receive a request from a client to access a server. The device can identify an index in a highest level. The device can determine a second index in the highest level that is after the index in the highest level and has a bit value indicating server availability. The device can identify an index in the lowest level mapping to the second index in the highest level. The device can select a server corresponding to the index in the lowest level.

Abstract: Systems, computer program products, and methods are described herein for implementing hybrid cloud computation engines using infrastructure optimization. The present invention may be configured to identify, from a group of available infrastructures, an infrastructure to perform a computing request, where the group of available infrastructures includes an enterprise infrastructure and one or more cloud provider infrastructures. The present invention may be further configured to provide, to the identified infrastructure and via a blockchain distributed ledger, the computing request and input data for the computing request, where the computing request includes a request to generate, using a recurrent neural network, based on data in the blockchain distributed ledger, and based on the input data, output data. The present invention may be further configured to receive, from the identified infrastructure and via the blockchain distributed ledger, the output data.

Abstract: Embodiments of the present disclosure may provide dynamic and fair assignment techniques for allocating resources on a demand basis. Assignment control may be separated into at least two components: a local component and a global component. Each component may have an active dialog with each other; the dialog may include two aspects: 1) a demand for computing resources, and 2) a total allowed number of computing resources. The global component may allocate resources from a pool of resources to different local components, and the local components in turn may assign their allocated resources to local competing requests. The allocation may also be throttled or limited at various levels.

Abstract: A method, computer program product, and computer system for identifying, by a computing device, a list of objects containing a plurality of physical layer blocks (PLBs). One or more next PLBs of the plurality of PLBs may be allocated from a selected free object of the list of objects. One or more additional free objects from the list of objects may be generated. Garbage collection may be performed between an inactive object of the plurality of objects and the selected free object.

Abstract: A scheduler includes circuitry configured to, based on similarity between execution time and power consumption information of jobs executed in a system, classifies jobs into groups, construct respective time series prediction models for the groups using a power waveform included in each of the groups as teacher data, predict a power waveform at an interval including a first time from each of the constructed time series prediction models, compare a power waveform at an interval including a first time of a job in execution for which power is to be predicted with the predicted power waveform of each of the groups to identify a similar time series prediction model, based on the identified time series prediction model, predict power consumption at a predetermined interval including a second time for the job for which power is to be predicted, and control job execution based on the predicted power consumption.

Abstract: Examples may include a method of instantiating a virtual machine; instantiating a virtual device to transmit data to and receive data from assigned resources of a shared physical device by receiving input data requesting assigned resources for the virtual device, allocating assigned resources to the virtual device based at least in part on the input data, and mapping a page location in an address space of the shared physical device for a selected one of the assigned resources to a page location in a memory-mapped input/output (MMIO) space of the virtual device; and assigning the virtual device to the virtual machine, the virtual machine to transmit data to and receive data from the physical device via the MMIO space of the virtual device.

Abstract: A mechanism is described for facilitating using of a shared local memory for register spilling/filling relating to graphics processors at computing devices. A method of embodiments, as described herein, includes reserving one or more spaces of a shared local memory (SLM) to perform one or more of spilling and filling relating to registers associated with a graphics processor of a computing device.

Abstract: Real time security, integrity, and reliability postures of operational (OT), information (IT), and security (ST) systems, as well as slower changing security and operational blueprint, policies, processes, and rules governing the enterprise security and business risk management process, dynamically evolve and adapt to domain, context, and situational awareness, as well as the controls implemented across the operational and information systems that are controlled. Embodiments of the invention are systematized and pervasively applied across interconnected, interdependent, and diverse operational, information, and security systems to mitigate system-wide business risk, to improve efficiency and effectiveness of business processes and to enhance security control which conventional perimeter, network, or host based control and protection schemes cannot successfully perform.

Abstract: Methods and systems for providing services using mixed instance catalogs are described herein. A catalog may comprise a plurality of first virtual machines and a plurality of second virtual machines. The capacity of a first virtual machine may be larger than the capacity of a second virtual machine. Connection requests to access a service associated with the catalog may be distributed among the plurality of first virtual machines and the plurality of second virtual machines.

Abstract: Methods, systems, and apparatuses for generating notifications corresponding to queries submitted for execution by virtual warehouses are described herein. A request to execute a query may be received. An execution plan, for the query, may be identified. A processing complexity for the query may be predicted based on the query and the execution plan. A notification may be generated based on the processing complexity meeting an alert threshold. A user device may display the notification. A response to the notification comprising a selection of a first virtual warehouse, of a plurality of virtual warehouses, to execute the query may be received.

Abstract: Embodiments of the present disclosure relate to a method for allocating computing resources, an electronic device, and a corresponding computer program product. The method may include: obtaining an available resource list associated with a computing resource requester according to determination that a resource use request from the computing resource requester is received, wherein the available resource list includes the number and available time periods of computing resources that can be provided by at least one computing resource provider. In addition, the method further includes: allocating computing resources to the computing resource requester based on the available resource list and the resource use request, so that the computing resource requester uses the allocated computing resources to run a workload. The embodiments of the present disclosure can flexibly allocate the computing resources, thereby realizing full utilization of the computing resources.

Abstract: Disclosed are systems, methods and computer-readable media for controlling and managing the identification and provisioning of resources within an on-demand center as well as the transfer of workload to the provisioned resources. One aspect involves creating a virtual private cluster within the on-demand center for the particular workload from a local environment. A method of managing resources between a local compute environment and an on-demand environment includes detecting an event associated with a local compute environment and based on the detected event, identifying information about the local environment, establishing communication with an on-demand compute environment and transmitting the information about the local environment to the on-demand compute environment, provisioning resources within the on-demand compute environment to substantially duplicate the local environment and transferring workload from the local-environment to the on-demand compute environment.

Abstract: A remote management system includes a device management service having multiple instances that performs parallel distributed processing through the instances and a worker service having multiple instances that performs parallel distributed processing through the instances. Each instance of the device management service acquires the status of processing related to the recovery of the relevant instance from counter information, and stores the status of the processing related to the recovery of the relevant instance in the counter information when having performed the processing. Each instance of the worker service acquires the status of processing related to the recovery of an instance of the device management service from the counter information, and stores the status of the processing related to the recovery of the instance of the device management service in the counter information when having performed the processing.

Abstract: An information processing apparatus including a plurality of groups, each group including a first memory, a second memory different in process speed from the first memory, and a processor including a memory controller that is connected to the first memory and the second memory and that controls an access from a process to the first memory and the second memory, wherein a first processor among a plurality of the processors of the plurality of groups is configured to determine, based on a characteristic of a plurality of the processes accessing data stored in the first memory or the second memory in each of the plurality of groups, an allocation of the plurality of processes onto the plurality of processors.

Abstract: Apparatus and associated methods relate to constructing a resource and attribute tasking solution to complete a user's objective with resource and attribute characteristics defining the tasked objects, in response to receiving a task definition, and satisfying a task definition constraint. An object includes of a set of resource and attribute dimensions each of which have a set of possible values. A tasking solution is the union of all valid solution vectors in the dimension space. The solution identifies all valid values to present to a user or other agent to make further decisions on further constraining resource and attribute characteristics for a final materializable tasking. In an illustrative example, the objective may be mapping resource supply to task demand. The task demand may be, for example, delivering a database-as-a-service platform configured based on rules generated to satisfy task definition capacity constraints.

Abstract: Wavefront loading in a processor is managed and includes monitoring a selected wavefront of a set of wavefronts. Reuse of memory access requests for the selected wavefront is counted. A cache hit rate in one or more caches of the processor is determined based on the counted reuse. Based on the cache hit rate, subsequent memory requests of other wavefronts of the set of wavefronts are modified by including a type of reuse of cache lines in requests to the caches. In the caches, storage of data in the caches is based on the type of reuse indicated by the subsequent memory access requests. Reused cache lines are protected by preventing cache line contents from being replaced by another cache line for a duration of processing the set of wavefronts. Caches are bypassed when streaming access requests are made.

Abstract: This disclosure describes techniques implemented partly by a service provider network for containerizing applications. In an example, the techniques may include requesting process relationship information for one or more potential processes of an application, receiving the requested process relationship information for the one or more potential processes of the application, and based on the received process relationship information, configuring a process relationship detection algorithm. Then, using the configured process relationship detection algorithm, the techniques may determine a respective relationship score for individual process pairs of processes operating on a system executing the application and determine one or more individual process pairs that have a respective relationship score that is equal to or above a threshold to be one or more cooperating process pairs.

Abstract: A computing device may include a memory and a processor cooperating with the memory and configured to receive connection leases providing instructions for connecting to computing sessions, and request connections to the computing sessions including the connection leases. Each connection lease may comprise a first component unique to a published resource, and a second component referenced by the first component and shared in common with a plurality of different published resources in other connection leases, with the second component being updateable independent of the first component.

Abstract: The present disclosure discloses a service migration method including: sending, by a VNFM module, a virtual machine VM request command to a first VIM module, where the first VIM module manages a first host on which post-upgrade new-version software is configured; receiving, by the VNFM module, a VM request response from the first VIM module, where the VM request response includes information about a first VM that the first VIM module requests on the first host, and virtual machine own data that can run on the first host and that is configured by the first VIM module is configured on the first VM; and sending, by the VNFM module, a service migration command to a virtualized network function VNF module, where the service migration command is used to instruct the VNF module to migrate a running service on a second VM to the first VM.

Abstract: A scoring platform may obtain a set of measurements associated with a service metric, wherein the service metric is associated with a service of a network. The scoring platform may determine, based on the set of measurements, an aggregation score associated with the service metric. The scoring platform may determine a scope score associated with the set of measurements, wherein the scope score is based on a quantity of units associated with the set of measurements. The scoring platform may determine a service impact score associated with the service metric based on the aggregation score and the scope score, wherein the service impact score is representative of a contributive effect associated with the service metric. The scoring platform may perform an action associated with the service impact score to permit a source of interest associated with the service metric to be detected.

Abstract: A query optimization system is described that, at runtime, optimizes the execution pipeline generated for a query. Based upon communications between nodes in the execution pipeline, the execution pipeline generated for a query is optimized by modifying the execution pipeline to create a modified execution pipeline. The modified execution pipeline is then executed to execute the query and results obtained for the query. The changes or modifications made to an execution pipeline may include changing the capabilities (e.g., changes to inputs and/or outputs of a node, changing the task(s) or function(s) performed by the node) of one or more nodes within the execution pipeline. The changes may include changing the position of one or more nodes within a directed acyclic graph representing the execution pipeline.

Abstract: The described technology is generally directed towards predicting a need for provisioned (e.g., cloud computing) resources so that the provisioned resources are proactively provisioned and operational before their actual need. Historical data is used to predict upcoming resource demand so that resources begin to be scaled up to meet demand in advance, at a prediction time, given that there is a “spin-up” delay between requesting resources and their actual availability. The predictive resource provisioning is adaptable to override customary historical data with expected exceptions, and is flexible to handle variable spin-up times, constraints, and optimizations.