Abstract: Disclosed herein are a distributed cloud system, a data processing method of the distributed cloud system, a storage medium. The data processing method of the distributed cloud system includes running an application of an edge computing system requested by a user device, generating a snapshot image of the application, and storing the generated snapshot image and transmitting the stored image during migration.

Abstract: A semiconductor device may include: first insulating pillars arranged in a first direction; second insulating pillars arranged alternately with the first insulating pillars and having a first width in the first direction and a second width in a second direction intersecting the first direction, the first width being greater than the second width; first memory cells located between the second insulating pillars and stacked along a first sidewall of each of the first insulating pillars; and second memory cells located between the second insulating pillars and stacked along a second sidewall of each of the first insulating pillars.

Abstract: The present invention relates to a compound capable of lowering the flammability of a non-aqueous electrolyte when included in the non-aqueous electrolyte and improving the life properties of a battery by forming an electrode-electrolyte interface which is stable at high temperatures and low in resistance, and relates to a compound represented by Formula I descried herein, a non-aqueous electrolyte solution and a lithium secondary battery both including the compound, n, m, Ak, and X are described herein.

Abstract: Disclosed herein are an intelligent scheduling apparatus and method. The intelligent scheduling apparatus includes one or more processors, and an execution memory for storing at least program that is executed by the one or more processors, wherein the at least one program is configured to, in a hybrid cloud environment including a cloud, an edge system, and a near-edge system, configure schedulers for scheduling tasks of the cloud, the edge system, and the near-edge systems, store data, requested by a client, in a work queue by controlling the schedulers based on a scheduler policy and process the tasks based on data stored in the work queue, and collect history data resulting from processing of the tasks depending on the scheduler policy, and train the scheduler policy based on the history data.

Abstract: Disclosed herein is a multi-cloud edge system. The multi-cloud edge system includes a core cloud, a multi-cluster-based first edge node system, and a multi-cluster-based near edge node system, wherein the multi-cluster-based first edge node system includes multiple worker nodes, and a master node including a scheduler.

Abstract: Disclosed herein are a distributed cloud system, a data processing method of a distributed cloud system, and a storage medium. The data processing method of a distributed cloud system includes receiving a request of a user for an edge cloud and controlling a distributed cloud system, wherein the distributed cloud system comprises a core cloud including a large-scale resource, the edge cloud, and a local cloud including a middle-scale resource between the core cloud and the edge cloud, processing tasks corresponding to the user request through a scheduler of the core cloud, distributing the tasks based on a queue, and aggregating results of processed tasks, and providing processed data in response to a request of the user, wherein the distributed cloud system provides a management function in case of failure in the distributed cloud system.

Abstract: Disclosed herein are an intelligent scheduling apparatus and method. The intelligent scheduling apparatus includes one or more processors, and an execution memory for storing at least program that is executed by the one or more processors, wherein the at least one program is configured to, in a hybrid cloud environment including a cloud, an edge system, and a near-edge system, configure schedulers for scheduling tasks of the cloud, the edge system, and the near-edge systems, store data, requested by a client, in a work queue by controlling the schedulers based on a scheduler policy and process the tasks based on data stored in the work queue, and collect history data resulting from processing of the tasks depending on the scheduler policy, and train the scheduler policy based on the history data.

Abstract: Disclosed herein is a method for resource allocation in an edge-computing environment. The method includes receiving a request for an intelligent edge service, selecting the worker server to execute the service based on an input/output congestion level, allocating resources based on topology information of the worker server, and configuring a virtual environment based on the allocated resources.

Abstract: Disclosed herein is a multi-cloud edge system. The multi-cloud edge system includes a core cloud, a multi-cluster-based first edge node system, and a multi-cluster-based near edge node system, wherein the multi-cluster-based first edge node system includes multiple worker nodes, and a master node including a scheduler.

Abstract: Disclosed herein are a Docker image creation apparatus and method. The Docker image creation apparatus may include one or more processors, and execution memory for storing at least one program that is executed by the one or more processors, wherein the at least one program is configured to create a Docker image based on a previous Docker file, and executing a push command to store the created the Docker file to a registry, wherein the Docker file is created from a partial Docker file of the previous Docker file, wherein the Docker image is created from a verified partial Docker file, after the partial Docker file is verified.

Abstract: Disclosed herein are an apparatus and method for managing in-memory container storage. The apparatus includes one or more processors, executable memory for storing at least one program executed by the one or more processors, and a container file system for storing a container, which provides application virtualization. Here, the container file system includes a merged access layer, a container layer, and an image layer, and the at least one program provides an application with link information of files in the container layer and the image layer, thereby allowing the application to access the files.

Abstract: Disclosed herein are an apparatus and method for managing memory-based integrated storage. The apparatus includes one or more processors and executable memory for storing at least one program executed by the one or more processors. The at least one program converts data operation tasks in response to a request for access to memory-based integrated storage from a user, a single virtual disk of a virtual storage pool of the memory-based integrated storage converts a disk access command into a command for connecting to a storage backend depending on the data operation tasks, and conversion of the data operation tasks into the command includes target identification indicating which local storage of the memory-based integrated storage is to be used.

Abstract: Disclosed herein are an apparatus and method for autoscaling a service shared in a cloud. The apparatus may include memory in which at least one program is recorded and a processor for executing the program, and the program may perform autoscaling by which at least one second service for performing the same function as a first service is additionally generated or deleted depending on a load that is incurred when multiple clients call the first service in the cloud. The at least one second service may be set to one of two or more execution types having different response times depending on a response time required by each of the multiple clients.

Abstract: Disclosed herein are an intelligent scheduling apparatus and method. The intelligent scheduling apparatus includes one or more processors, and an execution memory for storing at least program that is executed by the one or more processors, wherein the at least one program is configured to, in a hybrid cloud environment including a cloud, an edge system, and a near-edge system, configure schedulers for scheduling tasks of the cloud, the edge system, and the near-edge systems, store data, requested by a client, in a work queue by controlling the schedulers based on a scheduler policy and process the tasks based on data stored in the work queue, and collect history data resulting from processing of the tasks depending on the scheduler policy, and train the scheduler policy based on the history data.

Abstract: A method and apparatus are disclosed which relate generally to worker-scheduling technology in a serverless cloud-computing environment, and more particularly, to technology that allocates workers for executing functions on a micro-function platform which provides a function-level micro-service. The method and apparatus process a worker allocation task in a distributed manner as two-step pre-allocation schemes before a worker allocation request occurs, and pre-allocates workers required for a service using a function request period and a function execution time, thus minimizing scheduling costs incurred by worker allocation requests.

Abstract: Disclosed herein are an apparatus and method for setting the allocation rate of a parallel-computing accelerator. The method includes monitoring the utilization rate of the parallel-computing accelerator by an application and setting a start point, at which measurement of utilization data to be used for setting the allocation rate of the parallel-computing accelerator for the application is started, using the result of monitoring the utilization rate; setting an end point, at which the measurement of the utilization data is finished, based on the monitoring result; and setting the allocation rate of the parallel-computing accelerator using the utilization data measured during a time period from the start point to the end point.

Abstract: An apparatus and method for executing a function. The apparatus includes one or more processors and executable memory for storing at least one program executed by the one or more processors, and the at least one program is configured to determine whether it is possible to reengineer a user function source using interface description language (IDL) code, to generate a reengineered function source by reengineering the user function source, and to execute the reengineered function source.

Abstract: Disclosed are a method, apparatus, and system for selectively providing a virtual machine through actual measurement of efficiency of power usage. When a user terminal requests to provide a virtual machine, candidate virtual machines are activated on multiple virtual machine servers. Input data provided by the user terminal are provided to each of the multiple candidate virtual machines through replication and network virtualization, and identical candidate virtual machines are run on the multiple virtual machine servers through replication and network virtualization. When the candidate virtual machines are run, one of the candidate virtual machines is finally selected as the virtual machine to be provided to the user terminal based on efficiency of power usage.

Abstract: Disclosed herein are an apparatus and method for setting the allocation rate of a parallel-computing accelerator. The method includes monitoring the utilization rate of the parallel-computing accelerator by an application and setting a start point, at which measurement of utilization data to be used for setting the allocation rate of the parallel-computing accelerator for the application is started, using the result of monitoring the utilization rate; setting an end point, at which the measurement of the utilization data is finished, based on the monitoring result; and setting the allocation rate of the parallel-computing accelerator using the utilization data measured during a time period from the start point to the end point.

Abstract: A worker-scheduling method in a cloud-computing system and an apparatus for the same. The worker-scheduling method includes performing a first load-distribution operation of pre-creating template workers so as to process worker execution preparation loads in a distributed manner before a worker allocation request for function execution occurs, predicting a number of workers to be pre-allocated in consideration of variation in a worker allocation request period for each function, and performing a second load distribution operation of pre-allocating ready workers by performing worker upscaling on as many template workers as the number of workers to be pre-allocated.