Abstract: Methods and systems for inference generation are disclosed. To manage inference generation, a system may include an inference model manager and any number of data processing systems. The inference model manager may partition an inference model into portions. Portions of the inference model may be distributed to data processing systems in accordance with an execution plan. The execution plan may include instructions for timely execution of the inference model with respect to the needs of a downstream consumer. The inference model manager may manage execution of the inference model by monitoring the functionality of the data processing systems and dynamically re-assigning and/or re-locating data processing systems in the event that one or more data processing systems becomes unable to execute a portion of the inference model.

Abstract: Techniques are provided for data-driven virtual machine restoration. In an example, a VM crashes and is to be restored. There can be multiple restoration paths that can be used to recover the VM (e.g., various source locations where a recovery image is stored, various recovery images, and various target locations where a VM can be restored to). A trained prediction model can analyze these various restoration paths and predict which restoration path will have a quickest time to recovery, to minimize a time that the VM is unavailable.

Abstract: A Function as a Service (FaaS) distribution system is configured to implement FaaS as a Service (FaaSaaS), enabling autonomous storage systems to be used as FaaS providers during periods where the storage systems are not being used at full capacity to process primary workloads. The FaaS distribution system receives functions from FaaS consumers, and selects a FaaS provider from a set of autonomous storage systems currently able to process FaaS workloads. The FaaS distribution system selects FaaS providers based on an expected execution time of the function and expected execution times of other functions executing on particular FaaS providers, to preferentially select a FaaS provider that is currently running an instance of the function, and to preferentially select a FaaS provider that has other functions that are current executing that are not expected to finish execution at the same time the current function is expected to complete execution.

Abstract: A method, computer program product, and computing system for processing historical input/output (IO) performance data associated with one or more storage objects of a storage system. A plurality of IO modeling systems may be trained using the historical IO performance data. Modeling performance information may be determined for the plurality of IO modeling systems across the historical IO performance data. A forecast score may be determined for each IO modeling system based on the modeling performance information for the plurality of IO modeling systems. A subset of the plurality of IO modeling systems may be selected based upon the forecast score for each IO modeling system. The at least one IO modeling system may be trained using the historical IO performance data. IO performance data may be forecasted using the at least one trained IO modeling system from the subset of the plurality of IO modeling systems.

Abstract: A method, computer program product, and computing system for processing historical input/output (IO) performance data associated with one or more storage objects of a storage system. A smoothing model may be applied on at least a portion of the historical IO performance data to generate forecast IO performance data. The forecast IO performance data may be compared to observed IO performance data to generate one or more performance differentials. A normal IO performance range may be generated based upon, at least in part, the one or more performance differentials. One or more IO performance anomalies may be detected based upon, at least in part, the normal IO performance range.

Abstract: An organizational graph and a microservice graph can be generated. For two neighboring microservices, a combined organizational distance can be determined based on a distance for respective workers of each microservice in the organizational graph. The combined organizational distance can be used to determine a priority for a connection between the two microservices, and chaos testing can be performed on that connection based on its priority.

Abstract: One example method includes collecting telemetry data for each of a group of virtual machines (VM), and each of the VMs is associated with a user, collecting usage data for each of the VMs, creating a user profile definition for each user, and the user profile definition is created based on the telemetry data and usage data of the VMs associated with that user, creating, for each user, a user profile that is based on the user profile definition for that user, clustering the users based on similarity of their respective user profiles, and generating a recommended VM hardware configuration for a VM of one of the users.

Abstract: A disclosed information handling system includes an edge device communicatively coupled to a cloud computing resource. The edge device is configured to respond to receiving, from an internet of things (IoT) unit, a numeric value for a parameter of interest by determining a compressed encoding for the numeric value in accordance with a non-lossless compression algorithm. The edge device transmits the compressed encoding of the numeric value to the cloud computing resource. The cloud computing resource includes a decoder communicatively coupled to the encoder and configured to respond to receiving the compressed encoding by generating a surrogate for the numeric value. The surrogate may be generated in accordance with a probability distribution applicable to the parameter of interest. The compression algorithm may be a clustering algorithm such as a k-means clustering algorithm.

Abstract: Co-clustering of at least some parameters is employed to reduce data transferred between edge and cloud resources. Single-parameter cluster information, including cluster counts, for each of two or more parameters of interest is accessed. Each parameter may represent a time series of numeric values sent from an IoT unit to an edge device. A co-clustering ratio is determined for each unique parameter pair. The co-clustering ratio indicates whether the number of clusters produced by a co-clustering algorithm applied to a group of parameters is less than the number of clusters required to represent the parameters without co-clustering. Co-cluster groups may be identified based on the cluster ratios. For each co-cluster group, the co-clustering algorithm may be invoked to produce compressed encodings of numeric value tuples. The compressed encoding is then transmitted to a cloud computing resource and decoded into a tuple of surrogate values.

Abstract: A Function as a Service (FaaS) distribution system is configured to implement FaaS as a Service (FaaSaaS), enabling autonomous storage systems to be used as FaaS providers during periods where the storage systems are not being used at full capacity to process primary workloads. The FaaS distribution system receives functions from FaaS consumers, and selects a FaaS provider from a set of autonomous storage systems currently able to process FaaS workloads. The FaaS distribution system selects FaaS providers based on an expected execution time of the function and expected execution times of other functions executing on particular FaaS providers, to preferentially select a FaaS provider that is currently running an instance of the function, and to preferentially select a FaaS provider that has other functions that are current executing that are not expected to finish execution at the same time the current function is expected to complete execution.

Abstract: Techniques are provided for adaptive sensor position determination for multiple mobile sensors.

Abstract: Techniques for detecting impactful performance anomalies in storage systems. The techniques include obtaining, for each performance metric of a storage system's workload, a training set of series diffs based on a threshold. Each diff represents a difference between an observed value from an observed set of time series values for the performance metric and a normalized value from a corresponding set of normalized time series values. The techniques include applying the training set of series diffs for each performance metric to an unsupervised anomaly detection algorithm and running the algorithm to identify potentially impactful anomalies in a multi-dimensional search space. The techniques include identifying impactful anomalies from among the potentially impactful anomalies that exceed an anomaly score.

Abstract: Techniques are used for balancing load on a storage system according to multiple variables. The techniques may be used to provide, among other things, defining, across at least two variables, a balance constraint for a load on a storage system. Among a set of transfers of volumes from one node to another node in the storage system, a transfer of a volume that minimizes the distance between the load and an ideal balanced state of the storage system is identified. The identified transfer of a volume is added to a combination of transfers of volumes. Whether the combination of transfers of volumes meets the balance constraint is determined. If the combination meets the balance constraint, the combination is selected as a solution to balance the load.

Abstract: Techniques are provided for customer support ticket aggregation. One method comprises obtaining a customer support ticket; extracting a topic of the customer support ticket using a topic model based on natural language processing techniques; converting the customer support ticket to a topic vector representation that identifies the extracted topic and comprises a list of words describing the topic based on a collection of processed customer support tickets; extracting features from the customer support ticket; generating a fingerprint for the customer support ticket that comprises the topic vector representation and the extracted features; applying the fingerprint to a machine learning similarity model that compares the fingerprint to fingerprints of processed customer support tickets from the collection of processed customer support tickets; and identifying a processed customer support ticket from the collection of processed customer support tickets that is related to the customer support ticket.

Abstract: One example method includes encoding data as a polysaccharide structure, synthesizing the polysaccharide structure to create polysaccharide storage media that comprises the data, and storing the polysaccharide storage media. The example method may also include receiving a read request directed to the polysaccharide storage media, mapping the polysaccharide structure to create a map in response to the read request, traversing the map of the polysaccharide structure to determine an X-base number, and obtaining the data by converting the X-base number to a binary form.

Abstract: A system can determine to produce a storage device for a user identity indicative of a user. The system can determine a number of extra disks to include with the storage device as part of the production, the extra disks enabling further storage capacity for the storage device beyond a specified storage capacity, the determining of the number of extra disks being based on data from a group of data, the group of data comprising first cost data representative of a first cost associated with including the second number of extra disks, probability data representative of a probability that the further storage capacity beyond the specified storage capacity will be requested during a defined time period after the production, and second cost data representative of a second cost associated with installing the second number of extra disks after the storage device has been delivered to the user site.

Abstract: One example method includes monitoring performance of an element of a runtime environment, where the monitoring includes collecting performance information concerning the element, analyzing the collected information, detecting, based on the analysis of the collected information, an anomaly in the performance of the element and, in response to detection of the anomaly, automatically marking a snapshot of the runtime environment element, and the marking of the snapshot overrides a retention policy applicable to the snapshot.

Abstract: A method for alleviating data poisoning in an edge computing resource includes receiving a numeric value from an Internet of Things (IoT) unit and associating the numeric value with a cluster selected from a plurality of clusters in accordance with a suitable clustering algorithm such as a k-means clustering algorithm. In at least some embodiments, the numeric value comprises a poisoned numeric value including an adversarial component injected by an adversary to negatively impact a trained model of a cloud-based artificial intelligence engine. Rather than permitting the injected adversarial component to corrupt the AI engine, a cluster with which the numeric value is associated is sampled in accordance with a probability distribution of the cluster to obtain a surrogate for the poisoned numeric value. The surrogate may then be provided as an input to an inference module of the AI engine to generate a prediction.

Abstract: Co-clustering of at least some parameters is employed to reduce data transferred between edge and cloud resources. Single-parameter cluster information, including cluster counts, for each of two or more parameters of interest is accessed. Each parameter may represent a time series of numeric values sent from an IoT unit to an edge device. A co-clustering ratio is determined for each unique parameter pair. The co-clustering ratio indicates whether the number of clusters produced by a co-clustering algorithm applied to a group of parameters is less than the number of clusters required to represent the parameters without co-clustering. Co-cluster groups may be identified based on the cluster ratios. For each co-cluster group, the co-clustering algorithm may be invoked to produce compressed encodings of numeric value tuples. The compressed encoding is then transmitted to a cloud computing resource and decoded into a tuple of surrogate values.

Abstract: A system can establish a primary file system on a block array of computer storage, wherein the block array comprises a range of addresses, wherein the primary file system is configured to address the range of addresses. The system can establish a shadowed file system on the block array, wherein the shadowed file system is configured to access portions of the block array that are unused by the primary file system. The system can, in response to receiving a request to write data to the primary file system, and in response to determining that an amount of the block array utilized by the primary file system is full, transfer a first portion of the block array utilized by the shadowed file system to the primary file system.