Abstract: A checkpoint server for containers is disclosed. The checkpoint server generates checkpoint images of running containers or of warmed-up containers. These checkpoint images are restored such that the order in which memory pages are accessed can be recorded or logged. During a restore operation to a host, the memory pages are transmitted in accordance with the page order log. The container can then begin serving requests before all of the memory pages have been transmitted to the host.

Abstract: Mechanisms of cloning containers to spawn offspring, orchestrate new containers on different execution environments, and enabling intra-container communication while maintaining parent-child relationships are disclosed.

Abstract: A software and/or data mobility platform. The mobility of software and data in an edge network is achieved by loading software and/or data on an edge node. The software and data are replicated or migrated to neighbor nodes and prepared for the device when the device switches nodes. As the device switched nodes, clean up or mobility operations are performed by replicating or migrating the software/data to new neighbor nodes and deleting or removing the software/data from nodes that are no longer considered to be neighbor nodes. The software is typically deployed to the mobility platform rather than directly to the nodes to allow developers to focus on their application rather than on the mobility of the application.

Abstract: One example method includes creating a manifest that specifies one or more requirements concerning execution of an application that resides at an end device in an N-tier configuration, identifying a workload that is associated with the application and executable at one or more edge stations of the N-tier configuration, gathering and evaluating network telemetry, orchestrating the workload based on the network telemetry and the manifest, scheduling performance of the workload at the one or more edge stations, and performing the workload at the one or more edge stations in accordance with the scheduling.

Abstract: A method for managing data includes obtaining a request for a machine learning (ML) pipeline selection from a client, wherein the request comprises a training dataset and a domain of the training dataset, and in response to the request: identifying a set of ML pipelines based on the domain, obtaining runtime statistics for the set of ML pipelines using the domain and at least a portion the training dataset, generating, using a user preference model, an ordering of the set of ML pipelines based on the runtime statistics and user preferences, and presenting the ordering, the runtime statistics, and a notification based on the ordering to the client.

Abstract: A method for managing data includes obtaining, by a database proxy server, a document manifest, wherein the document manifest specifies a plurality of queries, wherein each of the plurality of queries specifies data and a policy to be implemented on the data, wherein the data is stored in a central database, and processing the plurality of queries.

Abstract: Techniques for dynamic application management are provided. For example, an apparatus comprises at least one processing platform configured to: execute a portion of an application program in a first virtual computing element, wherein the application program comprises at least one portion of marked code; receive a request for execution of the portion of marked code; determine, based at least in part on the portion of marked code, one or more cloud platforms on which to execute the portion of marked code; and cause the portion of marked code identified in the request to be executed on the one or more cloud platforms.

Abstract: One example method includes transmitting a request for a container image to a registry, receiving metadata associated with the container image, wherein the metadata allows a controller to mount an empty filesystem on a host machine, starting a container from the container image without receiving all files associated with the container image, receiving files, from a container server, needed by the container based on an access sequence associated with the container. This allows a container to be started without downloading the entire container image and also conversed bandwidth by providing the files as needed based on the manner in which the container accesses files during execution.

Abstract: A software and/or data mobility platform. The mobility of software and data in an edge network is achieved by loading software and/or data on an edge node. The software and data are replicated or migrated to neighbor nodes and prepared for the device when the device switches nodes. As the device switched nodes, clean up or mobility operations are performed by replicating or migrating the software/data to new neighbor nodes and deleting or removing the software/data from nodes that are no longer considered to be neighbor nodes. The software is typically deployed to the mobility platform rather than directly to the nodes to allow developers to focus on their application rather than on the mobility of the application.

Abstract: One example method includes receiving, as an input, an ML pipeline definition, identifying a group of layers required to be created for the ML pipeline definition, for one of more of the layers, receiving input concerning one or more characteristics of the layer, creating the layers for which input has been received, and packaging the created layers with the ML pipeline definition to create a production-ready ML model.

Abstract: One example method includes connecting to a server component, transmitting, to the server component, information concerning a hardware configuration associated with an asset having a capability that is fully utilized during a first time period and the capability is idle during a second time period, receiving, from the server component, cluster connection information, and using the cluster connection information to temporarily connect the asset to the cluster as a node of the cluster so that the capability is available during idle time to perform a workload of the cluster.

Abstract: One example method includes connecting to a server component, transmitting, to the server component, information concerning a hardware configuration associated with an asset having a capability that is fully utilized during a first time period and the capability is idle during a second time period, receiving, from the server component, cluster connection information, and using the cluster connection information to temporarily connect the asset to the cluster as a node of the cluster so that the capability is available during idle time to perform a workload of the cluster.

Abstract: One example method, which may be performed at an end device configured to communicate with an edge station, includes listening for a broadcast signal from the edge station, joining a broadcast channel, receiving edge station information, selecting an edge station, transmitting a manifest to the selected edge station, receiving route information from the selected edge station, accessing a container identified in the route information, and issuing a call to the selected edge station to execute an application workload on the container.

Abstract: Techniques for function execution environment selection for a decomposed application are provided. In one example, an apparatus comprises at least one processing platform configured to execute a portion of an application program in a first virtual computing element, wherein the application program comprises one or more portions of marked code, receive a request for execution of one of the one or more portions of marked code, decide whether to execute the portion of marked code identified in the request in the first virtual computing element or in a second virtual computing element, determine an execution environment from one or more execution environments specified in the marked code for the second virtual computing element to execute the marked code, when it is decided to execute the portion of the marked code in the second virtual computing element, and cause the portion of marked code identified in the request to be executed.

Abstract: Techniques for dynamic decomposition of an application executing in a computing environment are provided. For example, an apparatus comprises at least one processing platform comprising one or more processing devices. The at least one processing platform is configured to execute a portion of an application program in a first virtual computing element, wherein the application program comprises one or more portions of marked code, receive a request for execution of one of the one or more portions of marked code, decide whether to execute the portion of marked code identified in the request in the first virtual computing element or in a second virtual computing element, and cause the portion of marked code identified in the request to be executed in the second virtual computing element, when it is decided to execute the portion of the marked code in the second virtual computing element.

Abstract: Techniques for dynamic decomposition of an application executing in a computing environment are provided. For example, an apparatus comprises at least one processing platform comprising one or more processing devices. The at least one processing platform is configured to execute a portion of an application program in a first virtual computing element, wherein the application program comprises one or more portions of marked code, receive a request for execution of one of the one or more portions of marked code, decide whether to execute the portion of marked code identified in the request in the first virtual computing element or in a second virtual computing element, and cause the portion of marked code identified in the request to be executed in the second virtual computing element, when it is decided to execute the portion of the marked code in the second virtual computing element.

Abstract: Methods, systems, and apparatuses for integrated circuit packages and lead frames are provided. A quad flat no-lead (QFN) package includes a plurality of peripherally positioned pins, a die-attach paddle, an integrated circuit die, and an encapsulating material. The die-attach paddle is positioned within a periphery formed by the pins. The die is attached to the die-attach paddle. The encapsulating material encapsulates the die on the die-attach paddle, encapsulates bond wires connected between the die and the pins, and fills a space between the pins and the die-attach paddle. One or more of the pins are extended. An extended pin may be elongated, L shaped, T shaped, or “wishbone” shaped. The extended pin(s) enable wire bonding of additional ground, power, and I/O (input/output) pads of the die in a manner that does not significantly increase QFN package cost.

Abstract: Methods, systems, and apparatuses for integrated circuit packages and lead frames are provided. A quad flat no-lead (QFN) package includes a plurality of peripherally positioned pins, a die-attach paddle, an integrated circuit die, and an encapsulating material. The die-attach paddle is positioned within a periphery formed by the pins. The die is attached to the die-attach paddle. The encapsulating material encapsulates the die on the die-attach paddle, encapsulates bond wires connected between the die and the pins, and fills a space between the pins and the die-attach paddle. One or more of the pins are extended. An extended pin may be elongated, L shaped, T shaped, or “wishbone” shaped. The extended pin(s) enable wire bonding of additional ground, power, and I/O (input/output) pads of the die in a manner that does not significantly increase QFN package cost.

Abstract: A transistor driver circuit with a plurality of transistors, each having source and drain regions formed in a substrate. At least first and second interconnect layers are formed on top of the substrate. A first plurality of contacts connect the source regions to one of the first or second interconnect layers. A second plurality of contacts connect the drain regions to the other of the first or second interconnect layers. The first and second interconnect layers cover a region above the substrate area in which the plurality of transistors reside so as to achieve a low ohmic result. The second interconnect layer has openings therein for one of the respective first or second plurality of contacts to pass therethrough and couple to the at least one first interconnect layer. Either the first or second interconnect layers can function as an input or output for the circuit.

Abstract: A transistor driver circuit with a plurality of transistors, each having source and drain regions formed in a substrate. At least first and second interconnect layers are formed on top of the substrate. A first plurality of contacts connect the source regions to one of the first or second interconnect layers. A second plurality of contacts connect the drain regions to the other of the first or second interconnect layers. The first and second interconnect layers cover a region above the substrate area in which the plurality of transistors reside so as to achieve a low ohmic result. The second interconnect layer has openings therein for one of the respective first or second plurality of contacts to pass therethrough and couple to the at least one first interconnect layer. Either the first or second interconnect layers can function as an input or output for the circuit.