Abstract: Techniques for processing a read I/O operation that reads first content stored at a target logical address can include: determining, using the target logical address as a first key to index into a first cache, whether the first cache includes a first cache entry caching first metadata used to access a first physical storage location including the first content stored at the target logical address; responsive to determining the first cache includes the first cache entry, determining, using the first metadata as a second key to index into a second cache, whether the second cache includes a second cache entry caching the first content stored at the target logical address; and responsive to determining the second cache includes the second entry, returning the first content from the second entry of the second cache in response to the read I/O operation.

Abstract: Techniques are disclosed for sparse layer-wise training of neural networks. An example system includes at least one processing device including a processor coupled to a memory. The at least one processing device can be configured to implement the following steps: obtaining class predictions while saving activations for only a number ‘k’ layers of a neural network, using the class predictions to calculate a layer shallowness measure for the neural network, using the layer shallowness measure to determine a number ‘u’ of layers to update in the neural network, and partially updating the neural network by training only the number ‘u’ layers of the neural network.

Abstract: A storage system is configured with pools of processor cores. Each pool corresponds uniquely to one of the supported service levels of the storage system. Processor cores within each pool run at a clock speed that is defined for the service level corresponding to the respective pool. Incoming IOs are enqueued for processing by cores of the pool corresponding to the service level of the storage group containing the storage object that is the target of the IO. Forecast IO demand and corresponding data access latency are computed for each service level. If the forecast latency of a service level includes values outside a compliance range, then cores are promoted to, or demoted from, the corresponding pool so that forecast latency stays within the compliance range, thereby reducing power consumption while promoting service level compliance.

Abstract: An information handling resource may include a circuit board comprising an electrically-conductive pad, a circuit package comprising an electrically-conductive pin, and reflowed solder electrically coupling the electrically-conductive pad to the electrically-conductive pin, the reflowed solder having embedded therein at least one bead configured to provide mechanical standoff between the pad and the pin.

Abstract: Techniques are disclosed for explainable federated learning. An example method includes receiving, at a central node, relative importances for a plurality of features input into a machine learning (ML) model usable at an edge node, thereby defining a plurality of feature importances, the central node being configured to communicate with the edge nodes; using, at the central node, an ML algorithm to classify the edge nodes into a number ‘k’ of node groups based on the feature importances; and for each node group among the ‘k’ node groups: generating, at the central node, an ML shared model using the feature importances associated with a selected subset of nodes in the node group; and deploying, at the central node, the shared model to each edge node in the node group.

Abstract: A container system facilitates selective migrations of data from high-wear inducing data devices to spare data devices to promote SSD wear levelling. Storage capacity is configured into same-size cells with each container having the same number of cells as RAID width W. The cells are distributed over W SSDs. The containers are subdivided into equal-size data devices, each distributed over the same set of W SSDs as its associated container. Candidate migration source-target pairs are selected using heuristics guided by a prediction model that correlates disk wear rate with total writes of existing workloads. Wear rate predictions are computed using a weighted sum of write rates of all data devices on a disk. Wear rate predictions for candidate migration pairs are computed inclusive of additional wear caused by data migration to select a migration plan.

Abstract: Systems and methods for loading firmware onto an external embedded controller (EC) of a heterogenous computing platform. In an illustrative, non-limiting embodiment, an Information Handling System (IHS) may include: a heterogeneous computing platform having a Reduced Instruction Set Computer (RISC) processor and a Peripheral Component Interconnect Express (PCIe) controller coupled thereto, and an EC external to the heterogeneous computing platform and coupled to the PCIe controller, wherein the EC is configured to retrieve at least one of: firmware instructions, a configuration setting, or a table, from a PCIe device coupled to the PCIe controller.

Abstract: One or more aspects of the present disclosure relate to cache layout optimization. In embodiments, an input/output (IO) workload is received by a storage array. Further, cache slot allocations are dynamically adjusted for each cache segment of global memory based on one or more characteristics of the IO workload.

Abstract: One or more aspects of the present disclosure relate to simulating an input/output (IO) workload. In embodiments, information corresponding to one or more characteristics of an input/output (IO) workload is received. In addition, metadata corresponding to the IO workload is generated using the information. Further, testing of a storage array can be enabled using the metadata.

Abstract: One or more aspects of the present disclosure relate to cache layout optimization. In embodiments, an input/output (IO) workload is received by a storage array. Additionally, a system-level cache layout distribution model is established based on one or more characteristics of the IO workload. Further, cache slot allocations are dynamically adjusted for each cache segment of global memory for each board based on the system-level cache layout distribution model and the one or more characteristics of the IO workload.

Abstract: An information handling resource may include a circuit board comprising an electrically-conductive pad and a dimple formed on the circuit board, the dimple having a depression surrounded by a raised perimeter raised above a surface of the circuit board.

Abstract: An information handling resource may include a circuit board comprising an electrically-conductive pad and a plated standoff plated onto a surface of the electrically-conductive pad.

Abstract: A method for forming a circuit board may include, in a conductive layer of the circuit board, applying a layer of metal foil to a first insulating layer of the circuit board, selectively plating the layer of metal with additional metal, removing portions of the layer of metal foil such that the selectively plating and removing steps create a plurality of conductive traces including a first conductive trace and a second conductive trace within a conductive layer of the circuit board, wherein a first thickness of the first conductive trace from a surface of the first insulating layer is different than a second thickness of the second conductive trace from the surface, and laminating a second insulating layer over the conductive layer.

Abstract: A load balancing method for use in conjunction with an application or service provided by a distributed computing system may begin by electing, from a group of participants, a leader for each of a plurality of tasks associated with the application or service. Responsive to detecting a signal or some other indication to run a particular task, the elected leader of the particular task may delegate responsibility to run the particular task to a particular participant. The particular participant, upon subsequently discovering that responsibility for the particular task has been delegated to it, responds by running the particular task. In some embodiments, the elected leader for a task may delegate responsibility for running the task to a least-loaded participant.

Abstract: Methods and systems for managing the operation of data processing systems are disclosed. A data processing system may include a computing device that may provide computer implemented services. To provide the computer implemented services, hardware components of the data processing system may need to operate in predetermined manners. To manage the operation of the hardware components, the data processing system may heat them when their temperatures fall outside of thermal operating ranges. To mitigate some risk associated with heating, the data processing system may proactively calibrate temperature sensors used to guide the heating process.

Abstract: Managing a lifecycle of data by identifying data objects that are subject to same control rules in each stage of the lifecycle as grouped data, where the control rules allow only authorized access to or authorized operations on the grouped data based on a current stage of the lifecycle. A dataset is generated for the grouped data by identifying metadata of the grouped data to be processed similarly within the lifecycle, and storing the metadata in the dataset. The control rules associated with the grouped data as stage tags for the dataset. Actions performed on the data referenced by the dataset are monitored to ensure that the monitored actions comply with control rules using the stage tags of the dataset.

Abstract: In general, embodiments relate to a method for backing up databases, the method includes receiving a backup request, wherein the backup request specifies a role of primary and an asset group identifier for an asset group, identifying, based on the backup request, a database in the asset group, and issuing the backup request for the database.

Abstract: A system that automatically determines ideal temporary backup targets (TBTs) to store and restore backup data. For multiple TBTs, backup data is saved with location information to restore incremental backup data from these multiple targets. Hashes of each file are stored in a Merkle tree to enable data restore workflows and minimizing an amount of storage required to hold backup data. The lowest tree level comprises hashes of the backup data chunks, and higher levels contain hashes of the next lower levels. A backup target location value is added to non-data chunk hash records in the higher levels to identify where any piece of backup data is located so that restore operations can occur when the backup data is spread across multiple backup targets.

Abstract: A first user-initiated data write request including stream data is received from an edge device executing in a first zone. The stream data is partitioned into at least one segment and an event included in the stream data is written into the at least one segment based on an event's routing key. A segment container of a segment store hosts the at least one segment. The stream data is transmitted to a tier-1 storage of the first zone in order to provide a low-latency access to the stream data for the first user. Thereafter, the stream data is moved to a tier-2 storage of a second zone for permanent storage of the stream data. When requested by a second user operating in the second zone, a read-only version of the segment container retrieves the stream data from the tier-2 storage.

Abstract: Applications can be automatically grouped in a secure workspace environment. An application organizer service can be deployed on a user computing device to intelligently group applications within secure workspaces based on resource requirements, peripheral usage, threat level and/or data sharing characteristics of the applications. The application organizer service may determine these characteristics in a variety of ways including through the monitoring of the applications as they execute inside the secure workspaces.