Abstract: An anomaly location identification device includes a determination unit configured to determine presence or absence of an anomaly by inputting part or all of information items output from a plurality of devices into an anomaly detection algorithm; a calculation unit configured to calculate, in response to a determination made by the determination unit that an anomaly is present, with respect to one of the information items, an index indicating a degree of contribution to the anomaly; and an identification unit configured to perform calculation by an analysis algorithm using a causal model receiving the index as input, to identify an anomalous device, to improve the precision and calculation speed related to identification of an anomaly location.

Abstract: Failure of a processing unit that processes a plurality of information pieces is discovered in a short time.

Abstract: An information processing apparatus manages backup data that can be used as installation data and receives a request from an external apparatus. Also, the present information processing apparatus, in a case where the received request is a request for acquiring backup data, generates, by using the backup data, the installation data to be provided based on information associated with the request origin and information associated with the acquisition origin of the backup data. Wherein the installation data is generated by deleting, as necessary, information to be concealed in the managed backup data.

Abstract: Cloud simulation or validation system allows for the simulation of a future node that may be deployed on a piece of hardware. The system may attempt to simulate the operating system for node-A on top of the hardware for node-A, including basic network connectivity. When a host is booted up with the simulated configuration, validation scripts may be run to verify that the site is correctly prepped for cloud deployment. With its pre-staged RAM-based OS temporarily loaded into the host's RAM memory, any set of OS-based scripts, tools or binaries, may be executed for simulation and validation based upon the intended role of the host onto which the cloud simulation or validation system configuration is loaded.

Abstract: A device for monitoring diagnostic commands to a control unit. The device includes an execution platform, and a security device connected to the execution platform, with a command filter and state machines. The execution platform is configured to generate the diagnostic commands based on predefined scripts. The command filter is configured to select valid commands from among the diagnostic commands, based on conditions following from states of the state machines. The security device is configured to relay the commands to the control unit.

Abstract: The present disclosure relates to generating an event specific log file for a computer system. The generation comprises storing buffered log data in at least one memory buffer during an execution of one or more threads in real time. The generation further comprises receiving thread execution data. The generation further comprises inputting the thread execution data into an anomalous program event detector module configured to generate an anomalous program event signal in response to detecting an anomalous program execution event. The generation further comprises searching the buffered log data for event specific log data in the at least one memory buffer with a log dependency analyzer module in response to the anomalous program event signal. The method further comprises writing the event specific log data to an event specific log file using a log archive writer module.

Abstract: Techniques are described for safely overwriting decided slots and in-order fault tolerant consensus logs for replicated services. Using techniques described herein, a broad class of already-existing consensus log protocols may be enhanced/extended to safely overwrite decided slots and provide in-order fault tolerant consensus logs. When changing to a different epoch of a consensus log, slots determined to be unreachable may be changed/deleted even if slots after the gap were decided. A sequencer protocol establishes distributed consensus among a group of services. The sequencer protocol provides in-order execution of messages from multiple clients, and flow control from within the sequencer protocol, without offloading de-duplicate and reorder (DDRO) logic to the application layer. Fault tolerance is provided by egress cursors and ingress cursors, which provide awareness of which specific messages from each client sender should be executed next, even if those messages are not presently in the consensus log.

Abstract: Examples described herein relate a method, a system, and a non-transitory machine-readable medium for restoring a computing resource. The method may include determining whether the computing resource is required to be restored on a recovery node using a backup of the computing resource stored in a backup storage node. A resource restore operation may be triggered on the recovery node in response to determining that the computing resource is required to be restored. The resource restore operation include copying a subset of the objects from the backup to the recovery node to form, from the subset of objects, a partial filesystem instance of the computing resource on the recovery node that is operable as a restored computing resource on the recovery node.

Abstract: According to one general aspect, an apparatus may include a storage element configured to store both data and metadata, wherein each piece of data is associated with and stored with a corresponding piece of metadata. The apparatus may include a controller processor. The controller processor may be configured to, in response to a piece of data being written to the apparatus: generate a piece of metadata that includes a set of parameters to facilitate a at least partial repair of a block information map, and embed the piece of metadata with the corresponding piece of data.

Abstract: A system and method for providing erasure code protection across multiple storage devices. A data switch in a storage system connects a plurality of storage devices to a remote host. Each storage device is also connected to a controller, e.g., a baseboard management controller. During normal operation, read and write commands from the remote host are sent to respective storage devices through the data switch. When a write command is executed, the storage device executing the command sends a copy of the data to the controller, which generates and stores erasure codes, e.g., on a storage device that is dedicated to the storage of erasure codes, and invisible to the remote host. When a device fails or is removed, the controller reconfigures the data switch to redirect all traffic addressed to the failed or absent storage device to the controller, and the controller responds to host commands in its stead.

Abstract: Techniques involve managing a storage system. In accordance with the techniques, a plurality of copies of metadata of the storage system are read from a plurality of storage devices in a resource pool of the storage system. The resource pool includes a first number of storage devices, and the metadata describes configuration information of the storage system. A second number of copies are selected from the plurality of copies based on version information in the plurality of copies, where the second number of copies comprises the metadata in the same version. It is determined whether a relation between the first number and the second number satisfies a predetermined condition. The second number of copies are identified as trusted metadata based on determining the relation satisfies the predetermined condition. With the foregoing example implementation, the metadata in the storage system may be managed with higher reliability.

Abstract: Example implementations relate to Basic Input Output System updates. In some examples, a computing device can include a memory, a processor to, in response to completion of a Basic Input/Output System (BIOS) update, generate and store a boot status variable in the memory, determine whether the BIOS update was successful, in response to the BIOS update being successful, delete the boot status variable from the memory and perform a power event, and in response to the BIOS update being unsuccessful, perform a BIOS recovery.

Abstract: A system and method for the distributed analysis of high frequency transaction trace data to constantly categorize incoming transaction data, identify relevant transaction categories, create per-category statistical reference and current data and perform statistical tests to identify transaction categories showing overall statistically relevant performance anomalies. The relevant transaction category detection considers both the relative transaction frequency of categories compared to the overall transaction frequency and the temporal stability of a transaction category over an observation duration. The statistical data generated for the anomaly tests contains next to data describing the overall performance of transactions of a category also data describing the transaction execution context, like the number of concurrently executed transactions or transaction load during an observation period.

Abstract: The present embodiments relate to data center outage detection and alert generation. An outage detection service as described herein can process near real-time data from various sources in a datacenter and process the data using a model to determine one or more projected sources of a detected outage. The model as described herein can include one or more machine learning models incorporating a series of rules to process near-real time data and offline data and determine one or more projected sources of an outage. An alert message can be generated to provide the projected sources of the outage and other data relevant to the outage.

Abstract: In a memory controller included in a memory system for processing a program operation fail, the memory controller controls a plurality of memory devices commonly coupled to a channel, the plurality of memory devices respectively performing preset program operations, and includes: a buffer memory for storing data to be stored in the plurality of memory devices, based on a buffer memory index; and a program error processor for acquiring fail data corresponding to a program operation fail from a fail memory device and acquiring reprogram data that is data to be stored together with the fail data, based on the buffer memory index.

Abstract: When barrier synchronization is executed between multiple nodes that include a node and execute processing for a same job, the node transmits first data indicating an intermediate result of processing of the node to another node with which processing of the node is to be synchronized first among the multiple nodes, and receives second data indicating an intermediate result of processing of the other node from the other node. The node stores the first data of the node in a first memory region of a memory provided for the node, and store the second data of the other node in a second memory region of the memory.

Abstract: In an example, a method includes analysing data collected from a service. A value representative of the number of anomalies in the data is generated, this value then being compared with a threshold. Depending on whether the value is greater or less than the threshold, a performance parameter of the monitoring service may be evaluated.

Abstract: Methods for determining a fixed point format for one or more layers of a DNN based on the portion of the output error of the DNN attributed to the fixed point formats of the different layers. Specifically, in the methods described herein the output error of a DNN attributable to the quantisation of the weights or input data values of each layer is determined using a Taylor approximation and the fixed point number format of one or more layers is adjusted based on the attribution. For example, where the fixed point number formats used by a DNN comprises an exponent and a mantissa bit length, the mantissa bit length of the layer allocated the lowest portion of the output error may be reduced, or the mantissa bit length of the layer allocated the highest portion of the output error may be increased. Such a method may be iteratively repeated to determine an optimum set of fixed point number formats for the layers of a DNN.

Abstract: A computer-implemented method is presented for detecting anomalies in dynamic datasets generated in a cloud computing environment. The method includes monitoring a plurality of cloud servers receiving a plurality of data points, employing a two-level clustering training module to generate micro-clusters from the plurality of data points, each of the micro-clusters representing a set of original data from the plurality of data points, employing a detecting module to detect normal data points, abnormal data points, and unknown data points from the plurality of data points via a detection model, employing an evolving module using a different evolving mechanism for each of the normal, abnormal, and unknown data points to evolve the detection model, and generating a system report displayed on a user interface, the system report summarizing the micro-cluster information.

Abstract: Transitioning from using a donor four site replication system to using a target four site replication system includes initiating synchronization between a first local storage system of the donor four site replication system that receives I/O from a host with an alternative first local storage system of the target four site replication system, the host switching to automatically fail over to a synchronous storage system of the target four site replication system that receives synchronous data from the alternative first local storage system following synchronizing the first local storage system with the alternative first local storage system, and the host switching to perform I/O operations with the alternative first local storage system following switching to automatically fail over to the synchronous storage system. The host may run the z/OS operating system. The host may automatically fail over to a storage system by switching subchannel sets.