Database capacity estimation for database sizing

    公开(公告)号:US11163735B2

    公开(公告)日:2021-11-02

    申请号:US16446967

    申请日:2019-06-20

    摘要: A method and system for sizing databases. Multiple patterns of historical capacity usage by multiple databases in a computer system are identified. The multiple patterns are statistically significant for multiple characteristics due to the multiple databases including an input value of each characteristic of multiple characteristics pertaining to utilization of the multiple databases. A weighted value of capacity usage is computed as ACP*FC*PC, wherein ACP=an average value of historical capacity points for each pattern, FC=fraction of the characteristics associated with each pattern, and PC=a fraction of the multiple databases used for each pattern. One or more current databases are sized via utilization of the computed total value of capacity usage. A virtual machine or a new server of the computer system is created. The virtual machine or new server includes the sized one or more current databases.

    Framework for authoring data loaders and data savers

    公开(公告)号:US11138031B2

    公开(公告)日:2021-10-05

    申请号:US16572176

    申请日:2019-09-16

    摘要: Implementing static loaders and savers for the transfer of local and distributed data containers to and from storage systems can be difficult because there are so many different configurations of output formats, data containers and storage systems. Described herein is an extensible componentized data transfer framework for performant and scalable authoring of data loaders and data savers. Abstracted local and distributed workflows drive selection of plug-ins that can be composed by the framework into particular local or distributed scenario loaders and savers. Reusability and code sparsity are maximized.

    SYSTEM AND METHOD FOR DEFENDING APPLICATIONS INVOKING ANONYMOUS FUNCTIONS

    公开(公告)号:US20210209227A1

    公开(公告)日:2021-07-08

    申请号:US17195038

    申请日:2021-03-08

    申请人: Twistlock, Ltd.

    摘要: A system and method for defending an application configured to invoke anonymous functions. The method includes analyzing the application to determine at least one branch of the application, wherein each branch is an instruction that deviates from a default behavior of the application; identifying a potential threat branch based on the at least one branch of the application and an anonymous function, the potential threat branch including a call to an anonymous function; and creating a secured instance of the application, wherein creating the secured instance of the application further comprises embedding a policy within the anonymous function of the identified potential threat branch.

    Neural network operation reordering for parallel execution

    公开(公告)号:US11016775B2

    公开(公告)日:2021-05-25

    申请号:US16453478

    申请日:2019-06-26

    摘要: Techniques are disclosed for reordering operations of a neural network to improve runtime efficiency. In some examples, a compiler receives a description of the neural network comprising a plurality of operations. The compiler may determine which execution engine of a plurality of execution engines is to perform each of the plurality of operations. The compiler may determine an order of performance associated with the plurality of operations. The compiler may identify a runtime inefficiency based on the order of performance and a hardware usage for each of the plurality of operations. An operation may be reordered to reduce the runtime inefficiency. Instructions may be compiled based on the plurality of operations, which include the reordered operation.

    Method and system for an end-to-end artificial intelligence workflow

    公开(公告)号:US10936969B2

    公开(公告)日:2021-03-02

    申请号:US15716385

    申请日:2017-09-26

    摘要: In general, certain embodiments of the present disclosure provide methods and systems for enabling a reproducible processing of machine learning models and scalable deployment on a distributed network. The method comprises building a machine learning model; training the machine learning model to produce a plurality of versions of the machine learning model; tracking the plurality of versions of the machine learning model to produce a change facilitator tool; sharing the change facilitator tool to one or more devices such that each device can reproduce the plurality of versions of the machine learning model; and generating a deployable version of the machine learning model through repeated training.

    Environment isolation method and device

    公开(公告)号:US10798218B2

    公开(公告)日:2020-10-06

    申请号:US16016389

    申请日:2018-06-22

    发明人: Songlin Zhang

    摘要: The present application discloses an environment isolation method. First, a pre-configured full address in a routing address pool in a Remote Procedure Call (RPC) client terminal is automatically refreshed according to a preset cycle. The RPC client terminal stops automatic refresh in the routing address pool after receiving an environment isolation instruction sent by an environment isolation device, and replaces a current address in the routing address pool with a combined address. When subsequently receiving a service call request sent by a user, the RPC client terminal generates a valid target address based on the combined address in the routing address pool, sends the service call request to an RPC server terminal corresponding to the target address, and returns to the user a service processing result returned by the RPC server terminal.

    FRAMEWORK FOR AUTHORING DATA LOADERS AND DATA SAVERS

    公开(公告)号:US20200012515A1

    公开(公告)日:2020-01-09

    申请号:US16572176

    申请日:2019-09-16

    摘要: Implementing static loaders and savers for the transfer of local and distributed data containers to and from storage systems can be difficult because there are so many different configurations of output formats, data containers and storage systems. Described herein is an extensible componentized data transfer framework for performant and scalable authoring of data loaders and data savers. Abstracted local and distributed workflows drive selection of plug-ins that can be composed by the framework into particular local or distributed scenario loaders and savers. Reusability and code sparsity are maximized.

    Framework for authoring data loaders and data savers

    公开(公告)号:US10445130B2

    公开(公告)日:2019-10-15

    申请号:US15617526

    申请日:2017-06-08

    摘要: Implementing static loaders and savers for the transfer of local and distributed data containers to and from storage systems can be difficult because there are so many different configurations of output formats, data containers and storage systems. Described herein is an extensible componentized data transfer framework for performant and scalable authoring of data loaders and data savers. Abstracted local and distributed workflows drive selection of plug-ins that can be composed by the framework into particular local or distributed scenario loaders and savers. Reusability and code sparsity are maximized.