Abstract: A method for optimizing resource allocation based on prediction with reinforcement learning includes the steps of: a) providing a prediction on the number of units of the resource for a workload in more than N timepoints after a 0-th timepoint to the processor; b) calculating at least one 0-th possible operation cost (POC) based on at least one possible provisioned number (PPN) at 1-th timepoint; c) repeating the following sub-steps for the i-th timepoint with i from 1 to N by the processor: c1) calculating at least one i-th possible operation cost (POCi); c2) finding out the smallest and the second smallest POCi; and c3) setting PPNi used to calculate the smallest POCi as an i-th assigned number; and d) provisioning 1 unit of the resource at the 0-th timepoint and i-th assigned number of units of the resource at the i-th timepoint for the workload by the processor.

Abstract: A prediction-based method for analyzing change impact on software components is disclosed.

Abstract: A method for establishing system resource prediction and resource management model through multi-layer correlations is provided. The method builds an estimation model by analyzing the relationship between a main application workload, resource usage of the main application, and resource usage of sub-application resources and prepares in advance the specific resources to meet future requirements. This multi-layer analysis, prediction, and management method is different from the prior arts, which only focus on single-level estimation and resource deployment. The present invention can utilize more interactive relationships at different layers to effectively perform predictions, thereby achieving the advantage of reducing hidden resource management costs when operating application services.

Abstract: A method for establishing system resource prediction and resource management model through multi-layer correlations is provided. The method builds an estimation model by analyzing the relationship between a main application workload, resource usage of the main application, and resource usage of sub-application resources and prepares in advance the specific resources to meet future requirements. This multi-layer analysis, prediction, and management method is different from the prior arts, which only focus on single-level estimation and resource deployment. The present invention can utilize more interactive relationships at different layers to effectively perform predictions, thereby achieving the advantage of reducing hidden resource management costs when operating application services.

Abstract: A method and a system for diagnosing remaining lifetime of storages in a data center are disclosed. The method includes the steps of: a) sequentially and periodically collecting operating attributes of failed storages along with time-to-fail records of the failed storages in a data center; b) grouping the operating attributes collected at the same time or fallen in a continuous period of time so that each group has the same number of operating attributes; c) sequentially marking a time tag for the groups of operating attributes; d) generating a trend model of remaining lifetime of the storages from the operating attributes and time-to-fail records by ML and/or DL algorithm(s) with the groups of operating attributes and time-to-fail records fed according to the order of the time tags; and e) inputting a set of operating attributes of a currently operating storage into the trend model to calculate a remaining lifetime therefor.

Abstract: A SSD caching system for hybrid storages is disclosed. The caching system for hybrid storages includes: a Solid State Drive (SSD) for storing cached data, separated into a Repeated Pattern Cache (RPC) area and a Dynamical Replaceable Cache (DRC) area; and a caching managing module, including: an Input/output (I/O) profiling unit, for detecting I/O requests for accesses of blocks in a Hard Disk Drive (HDD) during a number of continuously detecting time intervals, and storing first data corresponding to first blocks being repeatedly accessed at least twice in individual continuously detecting time intervals to the RPC area sequentially; and a hot data searching unit, for detecting I/O requests for accesses of a HDD during a independently detecting time interval, and storing second data corresponding to second blocks being accessed at least twice in the independently detecting time interval to the DRC area sequentially.

Abstract: A method and a system for diagnosing remaining lifetime of storages in a data center are disclosed. The method includes the steps of: a) sequentially and periodically collecting operating attributes of failed storages along with time-to-fail records of the failed storages in a data center; b) grouping the operating attributes collected at the same time or fallen in a continuous period of time so that each group has the same number of operating attributes; c) sequentially marking a time tag for the groups of operating attributes; d) generating a trend model of remaining lifetime of the storages from the operating attributes and time-to-fail records by ML and/or DL algorithm(s) with the groups of operating attributes and time-to-fail records fed according to the order of the time tags; and e) inputting a set of operating attributes of a currently operating storage into the trend model to calculate a remaining lifetime therefor.

Abstract: A system for extending life expectancy of disks in a cloud-based service system and the system using the method are disclosed. The present invention uses LSTM modeling and the k-means clustering algorithms to find out performance limit and target lifespan for a cluster of disks assigned for a specific workload running over the cloud-based service system. The disks can be predicted to have a minimum lifetime and the requirement of the workload can be satisfied. Meanwhile, the minimum lifetime is the longest life the disk can last.

Abstract: A method for achieving distributed deduplication for a storage system for Internet Of Things (IOT) backup in a data center and associated storage system are provided. The system includes a number of storage units. Each storage unit includes a number of to-be-stored-destinations; a control unit, for controlling operations of the storage unit; and a distributed deduplication module, for providing or updating the deterministic function to the control unit and the edge component, and executing each step of the method in the control unit and/or the edge component.

Abstract: A method for data protection in a cloud-based service system is disclosed. The method includes the steps of: A. collecting historical operating data of storage devices in the cloud-based service system; B. building up a life expectancy model and a next-7-days failure probability model by the collected operating data; C. inputting operating data in the past 24-hours into the life expectancy model and the next-7-days failure probability model for every storage device to obtain ranges of expected lifespans in respective groups and corresponding failure probabilities; and D. backing up data in the storage devices according to the results of step C.

Abstract: A method for optimizing storage configuration for future demand and a system applying the method are disclosed. The system includes a monitoring module, a storage recording module, a traffic modeling module, a rule-based decision module, and a storage management module. With performance values and utilization values provided from the monitoring module, a traffic status of data access in a particular time in the future can be generated. Then, a storage configuration with the workload requirement according to some rules can be available. The storage configuration is implemented to fulfill the requirement of the traffic status of data access.

Abstract: A method for deploying storage system resources with learning of workloads applied to a storage system is disclosed. The method utilizes state-action fuzzy rules and action-reward fuzzy rules for the storage system to learn a best dynamic deployment of resources from the parameters of workloads. No more human interference is required. It reduces work loading of the storage system administrator and enhances performance of the storage system.

Abstract: A system for extending life expectancy of disks in a cloud-based service system and the system using the method are disclosed. The present invention uses LSTM modeling and the k-means clustering algorithms to find out performance limit and target lifespan for a cluster of disks assigned for a specific workload running over the cloud-based service system. The disks can be predicted to have a minimum lifetime and the requirement of the workload can be satisfied. Meanwhile, the minimum lifetime is the longest life the disk can last.

Abstract: A method for determining data in cache memory of a cloud storage architecture and a cloud storage system using the method are disclosed. The method includes the steps of: A. recording transactions from cache memory of a cloud storage during a period of time in the past, wherein each transaction comprises a time of recording, or a time of recording and cached data been accessed during the period of time in the past; B. assigning a specific time in the future; C. calculating a time-associated confidence for every cached data from the transactions based on a reference time; D. ranking the time-associated confidences; and E. providing the cached data with higher time-associated confidence in the catch memory, and removing the cached data in the cache memory with lower time-associated confidence when the cache memory is full before the specific time in the future.

Abstract: A method and system for storage traffic modeling in a Software Defined Storage (SDS) is disclosed. The method includes the steps of: collecting observed values of at least one performance parameter in a period of time from a storage node; learning a trend structure of the at least one performance parameter varying with time from the observed values; and providing a predicted value of one performance parameter in a particular point in time in the future. The storage node is operated by SDS software. The trend structure is adjusted based on observed values collected after the period of time. The predicted value is an output of the trend structure which has been adjusted.

Abstract: A method for data protection in a cloud-based service system is disclosed. The method includes the steps of: A. collecting historical operating data of storage devices in the cloud-based service system; B. building up a life expectancy model and a next-7-days failure probability model by the collected operating data; C. inputting operating data in the past 24-hours into the life expectancy model and the next-7-days failure probability model for every storage device to obtain ranges of expected lifespans in respective groups and corresponding failure probabilities; and D. backing up data in the storage devices according to the results of step C.

Abstract: Methods for optimizing utilization of workload-consumed resources for time-inflexible workloads are disclosed. By sorting workload-consumed resource profiles representing individual workloads in one system according to an order of standard deviation or descending volume, two workload-consumed resource profiles can be combined to check if combination criteria are fulfilled. If any combination satisfies the combination criteria, corresponding workloads can be combined to share the same resource from the system. Thus, optimizing utilization of the workload-consumed resource can be achieved.

Abstract: A storage system having nodes with light weight containers is disclosed by the present invention. The storage system includes: at least one cluster, having a number of nodes linked to one another, each node having: a node host, operated by a storage operating system, and installed with a number of service containers, which are used for providing specific services to clients and/or operations of the node out of the service containers; and at least one node storage device, operated by the storage operating system, for providing storing and accessing of data for the service containers; and a messaging module, linked to each node host, for bridging the node hosts and/or node hosts in different clusters. Coarse-grained REST APIs are used for communication between any two service containers to reduce chatty communication in the storage system.

Abstract: Methods for optimizing utilization of workload-consumed resources for time-inflexible workloads are disclosed. By sorting workload-consumed resource profiles representing individual workloads in one system according to an order of standard deviation or descending volume, two workload-consumed resource profiles can be combined to check if combination criteria are fulfilled. If any combination satisfies the combination criteria, corresponding workloads can be combined to share the same resource from the system. Thus, optimizing utilization of the workload-consumed resource can be achieved.

Abstract: A method for deploying storage system resources with learning of workloads applied to a storage system is disclosed. The method utilizes state-action fuzzy rules and action-reward fuzzy rules for the storage system to learn a best dynamic deployment of resources from the parameters of workloads. No more human interference is required. It reduces work loading of the storage system administrator and enhances performance of the storage system.