115results about How to "Realize dynamic expansion" patented technology

The invention relates to a modeling system and method for a visual machine learning training model. The modeling system for a visual machine learning training model includes a process designer, a process analyzer, and a process scheduler, wherein the process designer is used for according to dragging operation of the graphic algorithm component selected by a user, establishing the data flow direction between the algorithms in the graphic algorithm component, and generating the process description language; the process analyzer is used for analyzing the process description language generated from the process designer, creating the corresponding learning component, and generating the corresponding Spark learning pipeline; and the process scheduler is used for submitting the Spark learning pipeline to a Spark cluster to perform model training. By selecting the corresponding graphic algorithm component, establishing the data flow direction between the algorithms through dragging, generating the process description language, analyzing the process description language, creating the corresponding learning components according to the node class name and the attribute, generating the corresponding Spark learning pipeline, and submitting the Spark learning pipeline to the Spark cluster for performing model training, the modeling system and method for a visual machine learning training model can realize high quality machine learning modeling.

The invention relates to a distributed metadata management method and system. The distributed metadata management method particularly includes the steps of carrying out storage: carrying out partition to form independent metadata nodes for storing metadata and independent user table nodes for storing user tables, adopting a plurality of metadata nodes to store multiple copies of the metadata, and forming main copy nodes and auxiliary copy nodes, wherein the same metadata are stored in the main copy nodes and the auxiliary copy nodes; carrying out verifying: carrying out data verification on the main copy nodes and the auxiliary copy nodes to guarantee that the metadata stored in the main copy nodes and the metadata stored in the auxiliary copy nodes are coincident; carrying out repairing: building monitoring rings based on the main copy nodes and the auxiliary copy nodes through the ZooKeeper technology, and when it is monitored that some main copy nodes or some auxiliary copy codes fail, triggering switching between the main copy nodes and the auxiliary copy nodes through the monitoring rings to achieve repairing of the failed nodes. The distributed metadata management system corresponds to the technical schemes of the distributed metadata management method in a one-to-one mode. The distributed metadata management method and system solve the problem of single-node failures in metadata management and achieves the aim of coincidence among multiple copies.

A unified access method for all service channels of electric power marketing relates to a channel access method. The traditional extensive customer service management mode cannot support personalized and precise novel customer service demands. The structures of all kinds of channels are complicated, which poses high security risks. The method comprises the following steps: conducting unified channel service access based on a distributed service application framework; and introducing a permission mechanism of registration, binding, real-name authentication and service password authentication to realize unified management of channel registration accounts and account permissions, including unified account management and unified permission management. Through the technical scheme, standardized access to all service channels of electric power marketing is completed. An electric power service card is built to realize unified account management and connectivity of service channels. Through registration, binding, service password authentication and real-name authentication, a unified permission authentication system for accounts is built. Therefore, the data consistency and operating precision of electric power marketing service channels are ensured.

The invention discloses a method and a device capable of realizing dynamic expansion of a WEB (World Wide Web) application resource. The method comprises the following steps of S1, building a virtual management cloud platform which can manage a plurality of physical servers; S2, building a virtual machine and mounting a load balancing server; S3, building a virtual machine and mounting WEB application servers; S4, allocating a load balancing server route on the WEB application servers; S5, dynamically expanding the number of the WEB application servers according to the answer number of a user request by the load balancing server. By the method and the device, when user visits are increased sharply, the number of virtual servers is dynamically and automatically expanded, and when the user visits are reduced, the number of virtual servers is reduced, so that dynamic expansion and effective utilization of the resource are realized.

The invention provides a method for automatically processing application in a PaaS platform. The method mainly comprises the following steps: the number and domain names of the instances deployed by the application are determined; when the application is uploaded to a shared memory of the PaaS platform, the PaaS system releases a message to all server resource pools; after receiving the message, the server on the PaaS platform judges whether the software environment and hardware resources required by the application are met and whether the instance is already deployed; if the server receiving the message meets the conditions required by the application, the server notifies the PaaS platform that the application can be deployed on the physical server and informs the application of the IP address and port for deployment through a message; and the PaaS platform obtains and analyzes N messages received at first according to the number N of the instances deployed according to the application, and deploys and starts the application instances on the target application server according to the message content.

The invention relates to a multilevel and multidimensional method and a device for analyzing data attributes, wherein the method comprises the following steps: building a public code platform for maintaining the public basic data, wherein the public basic data comprises basic data and/or a public code table; building a data relationship model for the basic data analysis as the basic data analysis platform; performing the multilevel and multidimensional analysis on the data attributes according to the data relationship of the basic data analysis platform. The multilevel and multidimensional method and the device for analyzing data attributes achieve a tree-shaped hierarchy show for the multidimensional analysis on the data attributes through the analysis dimensionalities of the basic data analysis types and the tree-shaped hierarchy expansion, and fast achieve the building and flexible expansion for the public basic data through the public code platform. The method and the device fast achieve the building of the data attribute analysis types and analysis dimensions based on the basic data analysis platform, and achieve the multilevel and multidimensional analysis on the data attributes and dynamic expansion of the analysis dimensions in a tree-shaped hierarchy expansion way. The dimensional data analysis rules are arranged to flexibly achieve the statistics summary of the analysis data.

The invention discloses an inventory and distribution management system based on cloud computing. A Hadoop software architecture is applied for achieving distributed processing of mass Web data of the inventory and distribution management system, so that the safety of the inventory and distribution management system with respect to mass data storage is improved; in the cloud computing technology, distributed data storage can achieve dynamic load balance and automatic takeover of a failure node, so that high reliability, high performance, high availability and high extendibility are achieved; the inventory and distribution management system is established on a cloud computing public technology service platform so that abundant enterprises on a supply chain can conduct inventory management and logistics distribution in a finer and dynamic mode, and informationized running cost of export-oriented enterprises is reduced; the system can be applied to intelligent terminals; by the adoption of a Redis database, the functions of cluster deployment dynamic expansion, real-time data analysis, real-time data query, real-time data collection, real-time uploading, real-time communication and the like can be achieved.

The invention discloses a storage optimization-based distributed graph processing method, and belongs to the field of graph calculation. The method comprises the following steps of: carrying out data partitioning in a data pre-processing stage; distributing graph partition data; starting data iteration processing; updating message transfer; making a work node extension decision; and ending the data processing. According to the method, a consistency hash algorithm is proposed to partition and store graph data, and an external storage mode-based distributed graph processing system is designed and realized; and by utilizing a dynamic storage optimization strategy, graph data processing load balance is realized according to partition storage of load adjustment graphs, so that the graph data processing speed is improved, the problem that the overall performance is reduced as hotspots are caused by load imbalance in the graph data processing process in the prior art is solved, and then the graph processing performance is improved.

The invention discloses and provides a dynamically constructed and autonomously managed mobile cloud storage cluster system, which provides and enhances the characteristics of availability, expansibility, resource utilization rate, maintainability and the like of a cloud storage system by dynamically constructing a mobile node system, dynamically entering or exiting backbone cloud storage cluster systems and realizing a certain degree of autonomous management and joint collaboration. The dynamically constructed and autonomously managed mobile cloud storage cluster system comprises the backbone cloud storage cluster systems and at least one mobile node system, wherein the interconnection between the two backbone cloud storage cluster systems, between the backbone cloud storage cluster systems and the mobile node system or between two mobile node systems can be realized by the conventional general or special network systems. By using the system, a distributed resource cluster platform is combined with a distributed application platform, so that the low-cost rental and application expansion are realized. The system can be applied to the field of cloud computation and cloud storage.

The invention discloses a method and a device for data storage. The method comprises the following steps of: assigning a first-level file folder for an established channel in a plurality of established first-level file folders; assigning a secondary-level file folder in a plurality of established secondary-level file folders under the first-level file folders; and establishing a file with a fixed size used for storing the data in the channel. In the invention, the form of assigning multilevel file folders for the channel realizes the data storage in the channel, which has no need to spend large amount of time for pre-establishing a file at the very start; and the form also realizes the dynamic addition and deletion, and the dynamic expansion or reduction of the capacity of the storage channel. Furthermore, by cycle overwriting of the files with fixed sizes, the invention completely avoids the generation of the file fragmentation, thus leading the storage performance of arrays or disks to have no reduction with the increasing of the service time.

The invention provides an information subscribing method, processing node equipment and bus equipment. The method comprises that at least one first processing node loaded on an information bus obtains information events corresponding to first configuration information from the information events according to first configuration information; the at least one first processing node processes the obtained information events according to the first configuration information, wherein the at least one first processing node is nodes loaded on the information bus in a plug-in mode dynamically. By the aid of the information subscribing method, dynamic extension of the bus can be achieved, and later maintenance of network security products with the information events serving as granularity is facilitated.

The invention provides a mass class locator data storage method. A mass class locator data storage system based on a MongoDB server cluster is built to serve as a supporting platform, a data fragment technology and a distributed type data storage technology are utilized, a sharing-nothing storage framework is adopted, data are divided according to the specific dimensionality needed by services and are averagely stored in all servers and disks, so that it is ensured that read and write loads of the system are balanced as much as possible, all IOs for data reading and writing are called, and the total IO supporting capacity of the cluster is obtained. In addition, a memory mapped file technology is utilized, accessed data information is stored in memories of the servers, and therefore performance delay possibly caused by the magnetic disk IOs is reduced to the maximum extent.

The invention belongs to the technological field of communication, and provides a method of achieving a function of network adapter unloading, a network adapter, a host system and a network device. When datagram received by the network adapter or the host system is an unknown datagram of the network adapter, the unknown datagram of the network adapter is analyzed to generate new strategy items which are updated into a host strategy table and a network adapter strategy table. For datagram containing new private or standard protocol header Tag, only a datagram analyzing function module is necessary to update, so that the datagram analyzing function module can process the datagram and issue the new strategy items into the host strategy table and the network adapter strategy table, but the network adapter is unnecessary to change, and the dynamic extension of the function of the network adapter unloading is achieved.

The invention relates to the technical field of middle stations, in particular to a micro-service-based middle station system and a software-defined middle station system construction method. The middle station system comprises: a reusable service component module used for abstracting reusable processing logics of a plurality of service systems; the configuration item module that is used for defining configuration items and realizing the configuration items; the configuration item synchronization component module that is respectively connected with the reusable service component module and theconfiguration item module, wherein the configuration item synchronization component module acquires configuration item definition and configuration item implementation, outputs the configuration itemdefinition and the configuration item implementation and acquires a configuration item set value, and the reusable service component module executes reusable processing logic and the configuration item implementation corresponding to the configuration item set value. According to the invention, the flexible control of the middle station system is realized, the stable operation of the system is ensured, the service innovation is realized based on the change of the service as required, the systematic cooperation, flexible operation, dynamic expansion and scenarized autonomy are realized, and the control is flexible and efficient.

The embodiment of the invention provides a method and apparatus for controlling virtual broadband access service, and a communication system. The method comprises the following steps: obtaining a performance parameter of sharing a protocol component in a resource pool in a virtual broadband access server; detecting a control operation satisfying a preset strategy according to the performance parameter, wherein the control operation is used for controlling the protocol component of the virtual broadband access server; and executing the detected control operation. When the method and apparatus in the embodiment of the invention are at work, the performance parameter of sharing the protocol component in the resource pool by the VBRS is obtained, for example, online user number, the memory utilization rate of the protocol component, the CPU utilization rate of the protocol component and the like, the enabling control is performed on the protocol component in the VBRS according to the performance parameter to control the volume expansion and volume retraction of the protocol component in the resource pool, dynamic volume expansion or volume retraction of the protocol component is achieved, and the resources of the protocol component can be used reasonably.

The invention aims to provide a method for deploying an image recognition service on a container cloud. The method comprises the following steps: after the basic service in the image recognition service is deployed, obtaining a deployment file corresponding to an image recognition business service in the image recognition service, wherein the business service in the image recognition business service comprises at least one of the following services: an image recognition management service, a front-end resource service, an index service, at least one recognition service, a feature extraction service, an account service, a content management service, a publishing service, an image recognition API service and a resource storage service; and for each business service in the image recognition business services, creating a container set on the container cloud according to the deployment file, starting the business service, loading a corresponding configuration file through the container setaccording to a configuration service address corresponding to the business service, and managing the business service through a container management platform.