267results about How to "High concurrency" patented technology

A data structure, added to a modified form of the Blink-tree data structure, tracks delete states for nodes. The index delete state (DX) indicates whether it is safe to directly access an index node without re-traversing the B-tree. The DX state is maintained globally, outside of the tree structure. The data delete state (DD) indicates whether it is safe to post an index term for a new leaf node. A DD state is maintained in each level 1 node for its leaf nodes. Delete states indicate whether a specific node has not been deleted, or whether it may have been deleted. Delete states are used to remove the necessity for atomic node splits and chains of latches for deletes, while not requiring retraversal. This property of not requiring a retraversal is exploited to simplify the tree modification operations.

The invention discloses a host monitoring system based on a microservice architecture. The system comprises a Jenkins continuous deployment environment module and a microservice module, wherein the Jenkins continuous deployment environment module is used for receiving upgrade codes of all microservices, conducting code update, testing and deployment on the microservices correspondingly and installing new microservices on a server; the microservices are saved in the microservice module, and each microservice is used for achieving one independent function on the server, providing functional services for a plug-in manager and providing an operation interface of the server for a system administrator; the plug-in manager is used for providing basic environments for operation of plug-ins and managing the plug-ins; the plug-ins are final execution units of the safety monitoring function of a terminal, and used for reading safety strategies issued by the server and managing and controlling the terminal according to description in the safety strategies.

The invention provides an e-commerce loan management method and system, and relates to the technical field of e-commerce. The e-commerce loan management method comprises that transaction data, behavior data, social data and personal information of a user are collected from multiple data sources, one or multiple thereof is / are selected as a main reference basis, and big-data analysis is combined with a risk control model to carry out user evaluation; operation data of multiple accounts of the data is obtained, and account association analysis is carried out on the accounts; and a credit score and limit of the user are evaluated according to the user evaluation condition and the account association analysis condition; according to a loan application of the user as well as the credit score and limit, a loan is provided for the user and a repayment plane is generated automatically. Thus, a loan product without mortgage and guarantee and using the credit degree of a borrower as a payment guarantee can be provided for the users, and operation development of merchants can be accelerated.

The invention relates to a method for testing concurrency property of a cloud platform based on federated research, and belongs to the technical field of testing of cloud platform property. The method comprises the following steps of obtaining the evaluation indexes of a cloud calculation platform based on federated research under the condition of different quantities of concurrent users, and evaluating the property of the cloud calculation platform; firstly, building a federated research system based on a cloud platform environment, wherein the federated research system which is built under the cloud platform environment sequentially comprises a physical resource layer, a virtual resource layer, a data accessing layer, an external service layer, a safety link layer, a business function layer and a system expression layer from bottom to top; respectively testing the transaction query response time, maximum amount of concurrent users and amount of transactions per second for transaction query of the cloud platform, so as to evaluate the concurrency property of the cloud platform. The method is used for testing the concurrency property of the cloud platform.

The invention discloses a distributed crawler architecture based on Kafka and Quartz and an implementation method thereof, and belongs to the technical field of computer data mining.The technical problem how to meet the requirement of a distributed crawler through cooperation of a stand-alone crawler architecture and a distributed tool is solved, and crawling queue message multi-node distribution and timing crawling are achieved.According to the technical scheme, the distributed crawler architecture based on Kafka and Quartz comprises a basic crawler assembly, a URL storage queue, a URL message distribution mechanism based on Kafka, a crawler operation scheduling mechanism based on Quartz and a front-end console; the implementation method of the distributed crawler architecture based on Kafka and Quartz comprises the following steps that parameters of a crawling inlet, a crawling rule, a crawling result storage way and a scheduling rule are set according to the page of the front-end console, and cluster nodes to be deployed are selected to be deployed.

The invention discloses a file uploading and searching method of a distributed file system, relates to the distributed file system and aims at providing a distributed file system based file uploading and searching method capable of realizing rapid positioning of an object file. The technical key points are that: a client sends out a file uploading request to a load balancing server; the load balancing server receives the file uploading request and generates a file storing name to ensure that the file name can directly locate the storage position of the file; the load balancing server distributes a file server used for storing the uploaded file for the client, and returns the file storing name and the address of the file server to the client; the client visits the file server based on the address, sends the file and the file storing name to the file server, and the like.

The invention provides a method and a system for optimization realization based on time dimension data real-time inquiry service. The method can effectively degrade the consistency of the time dimension metadata to the ultimate consistency through TSD-based time dimension data inquiry-based distributed design, and through relevant optimization based on the design including fragmentation and downsampling, so as to support arbitrarily horizontal expansion of the dimension data; meanwhile, the inquiry service can support real-time inquiry pressure of high concurrency and high throughput, so as to realize high availability and high expansibility of the time dimension inquiry service.

The embodiments of the present invention provide a method for data transmission, a device for data transmission and a computer readable storage medium. According to the method for data transmission, the device for data transmission and the computer readable storage medium, a producer agent reads message data sent by a producer from a first inter-process message queue; the producer agent sends themessage data to the redis; and a consumer agent reads the message data from the redis, and writes the message data into a second inter-process message queue, so that a consumer can read the message data from the second inter-process message queue. With the method, device and computer readable storage medium adopted, producers and consumers do not need network interaction, and high-performance, high-concurrency and highly-available message middleware is realized on the basis of the redis.

The invention discloses a neural network accelerator suitable for edge equipment and a neural network acceleration calculation method, and relates to the technical field of neural networks. The network accelerator comprises a configuration unit, a data buffer unit, a processing matrix component (PMs) and a post-processing unit, and a main controller writes feature parameters of different types ofnetwork layers into a register of the configuration unit to control the mapping of different network layer operation logics to the processing matrix hardware, so as to realize the multiplexing of theprocessing matrix component, i.e., the operation acceleration of different types of network layers in the neural network is realized by using one hardware circuit without additional hardware resources; and the different types of network layers comprise a standard convolution layer and a pooling network layer. The multiplexing accelerator provided by the invention not only ensures the realization of the same function, but also has the advantages of less hardware resource consumption, higher hardware multiplexing rate, lower power consumption, high concurrency, high multiplexing characteristic and strong structural expansibility.

The invention relates to the field of data management and discloses a block chain technology-based health detection data management system. The system comprises a plurality of terminal nodes, a plurality of super nodes, at least one user group and a root node server, wherein each terminal node corresponds to a service terminal device for bearing user health detection; the service terminal device is used for generating health detection data and caching the health detection data; the user group includes at least one super node; the root node server is used for converting the terminal nodes to the super nodes according to running time of the terminal nodes and system resource abundance degree, and establishing the user group; and the root node server is also used for synchronizing the health detection data stored between the super nodes in the user group. The invention also discloses a block chain technology-based health detection data management method. According to the block chain technology-based health detection data management system and the block chain technology-based health detection data management method, the safety, completeness and reliability of the user health detection data which are stored in a distributed system for a long time can be effectively guaranteed.

The invention provides a distributed high concurrent real-time message pushing method and a distributed high concurrent real-time message pushing system. The method comprises the steps of building a message connection between a first user and a web server, receiving a message sent by a second user to the web server after the second user logs in a web, verifying whether the message is valid, if no,ending an illegal access, if yes, performing high speed concurrent treatment on redis data, and performing kafka partition message treatment; synchronously consuming a kafka message via the web servers, and synchronously pushing same data via the multiple web servers to corresponding grouped users of a websocket connection, and receiving the message sent by the web server via the first user. According to the method and system provided by the invention, the retardance of the websocket instant messaging under the high concurrent scene can be reduced, and the expandability of the same can be improved.

The invention provides a method for processing a remote procedure call request. In the method, the procedure of processing the remote procedure call (RPC) request from a client side is divided into a plurality of flow segments, each flow segment comprises a plurality of threads, each thread in each flow segment processes the same subtask, and different flow segments process different subtasks. The tasks processed by the threads in each flow segment are simple, and for the request of each RPC client side, the response speed is quick, and thus the overall response capability of a server for processing the RPC request is improved. The resource can be allocated for each flow segment by dynamically regulating the number of threads in each flow segment, so that the load and the processing capability of each flow segment can be balanced. Therefore, various resources, including the processing capability, the storage resource, the network links and the like, of a computer system where an RPC server is arranged are fully used, and the requirements of a high-concurrency and high-throughput environment are satisfied.

The invention discloses a suspect early warning method based on face recognition. The suspect face image eigenvalue is calculated by extracting the suspect face image. The eigenvalues of suspects' face images are compared with those in the blacklist database filed by the public security bureau. Judging whether the characteristic value of the suspect's face image is recorded in the blacklist database filed by the public security bureau, if so, sending an alarm signal to the driver, uploading the alarm signal to the dispatching monitoring center, and sending the alarm information to the public security bureau; if not, the eigenvalue comparison analysis is performed again. The invention also discloses a suspect early warning system based on face recognition, comprising a face image extractionmodule, an eigenvalue comparison module, a suspect judgment module and a monitoring alarm module. The invention realizes high-precision and high-efficiency face recognition, facilitates the identification of suspect basic information, understands suspect personnel activity, and can real-time call monitoring video, on-line video-on-demand, saving bandwidth resources.

The invention relates to a cross-chain transaction method and device on a block chain and a storage medium, and belongs to the technical field of block chains, and the method comprises the steps: monitoring a second contract account on a third-party public chain through at least two first contract accounts on a mole chain; when the mole chain monitors that the second contract account receives a transaction request sent by the first external account, creating a second transaction resource equivalent to the first transaction resource indicated by the transaction request on a target contract account of the mole chain; the target contract account transfers the second transaction resource to a second external account corresponding to the mole chain; when the second external account initiates atransaction request, the second external account transfers the second transaction resource to the target contract account; Triggering the second contract account to carry out transaction according tothe transaction record on the target contract account; the technical problems that an existing cross-chain transaction needs to consume extra transaction expenses and the transaction time is long canbe solved. The transaction cost of the cross-chain transaction is saved, and the transaction duration of the cross-chain transaction is shortened.

The invention relates to a multi-version GIS slicing service system which comprises a file storage server, an application cache server, a slicing server and a client, wherein the file storage server is used for storing slicing associated files in layers according to types and versions, wherein the slicing associated files include a slicing data file, a slicing resource description file and a slicing index file, the slicing data file is used for storing slicing data information, the slicing resource description file is used for describing the type, grade, coordinate and version information of the slicing data, and the slicing index file is used for describing the index information of the slicing data; the application cache server is used for loading the slicing index information into the application cache server; the slicing server is provided with an access interface capable of receiving a slicing request and is used for receiving a slicing request of the client, acquiring the corresponding slicing index information and slicing data and returning the corresponding slicing index information and slicing data to the client; the client is used for sending the slicing request and receiving the returned slicing data. The system is capable of providing access and management for multiple versions of GIS slicing, and is simple in system structure and good in use effect.

The invention discloses an automatic real-time supervision reporting system based on a double-chain architecture blockchain. The automatic real-time supervision reporting system comprises the following forming parts: a TBC / ABC double-chain architecture blockchain system, a supervision law and regulation library, and a supervision executing engine. The system is capable of using a TBC (Trading Blockchain) and ABC (Account Blockchain) double-chain architecture to respectively store trading information and account information, and using a TR (Trading Record) technology, realizing the high-concurrency high-speed automatic real-time supervision.

The invention provides a continuous sequence number generation system and a continuous sequence number generation method. The system comprises N servers, wherein the N servers are mutually connected, each server comprises a sequence number maintenance file and a reception buffer memory file, the N is a positive integer, the sequence number maintenance file is used for storing present largest sequence numbers, the reception buffer memory file is used for receiving synchronization signals sent by other servers, the synchronization signals comprise present largest sequence numbers of other servers, after a sequence number distribution request is received by the Mth server, the M is greater than zero and is smaller than or equal to the N, the Mth server is used for acquiring the present largest sequence number of the Mth server; a new sequence number is generated according to the present largest sequence numbers, response feedback is further distributed according to the sequence number, the synchronization signals are sent to other servers by the Mth server, and the synchronization signals have the new sequence number. The continuous sequence number generation system realizes high-efficiency and reliable generation of continuous sequence numbers in a distributed environment and improves availability.

The invention relates to a logging method used in a distributed environment, and the method comprises the following steps: configuring a log server and a log strategy of a client; sending logs of the client to the log server in real time; and storing the logs received by the log server according to the log strategy. According to the logging method disclosed by the invention, high-concurrency and high-capacity log services can be provided. Simultaneously, a system is easy to expand and can meet various different requirements. The logging method disclosed by the invention has no limitations on language of the client and can meet the requirements of various environments. In addition, the bandwidth can be fully utilized by using the logging method.

The invention relates to a method and a system for showing and querying logs in real time. The method comprises steps as follows: log messages are collected by a LOG collecting server Logstash and sent to a message system server Kafka in a message mode; a Tornado server reads log message records from the message system Kafka in a multi-thread concurrent reading mode and provides filter service; the Tornado server adopts a high-performance Websocket long link to send log messages to a foreground terminal Livetail quickly; the foreground terminal Livetail displays the log messages in real time in a specific manner and provides filtering and searching functions. The collected log messages are directly written into the storage management system Kafka, then read by the Tornado concurrently and sent to the front-end server Livetail with data filtering and querying functions by using efficient WebSocket, so that information time delay is reduced, the timeliness of the log messages is greatly improved, a filtering mechanism is provided at the server end, the messages are effectively pushed, the safety of the log messages is guaranteed, defects of existing systems are greatly overcome, and the user experience is improved.

The invention provides a sparse neural network processor based on a systolic array. The sparse neural network processor comprises a storage unit, a control unit, a sparse matrix operation array, a calculation unit and a confluence array. The storage unit is used for storing weights, gradients, features and instruction sequences used for scheduling data streams. The control unit takes out data required by the training and reasoning process from the storage unit according to the control of the instruction sequence, converts the data into a sparse matrix operation format and sends the data into the sparse matrix operation array. The sparse matrix operation array comprises a plurality of processing units connected in a systolic array mode and is used for completing sparse matrix operation. Thecalculation unit is used for completing element-by-element operation such as a nonlinear activation function. The confluence array delivers the same data segment to different rows of the systolic array through internal data transfer to reduce storage overhead. The processor makes full use of the sparsity of the weight and the characteristics, achieves the improvement of the speed and power consumption ratio in the neural network training and reasoning process, and has the advantages of high concurrency, low bandwidth requirements and the like.

A method for achieving an NoSQL key-value storage engine comprises two stages: firstly obtaining NoSQL storage engine Key Value SE: A1, removing an independent module from A relational database management system (RDBMS), and only keeping the storage engine in the RDBMS; A2, establishing a single-row byte string data line structure serving as a data line structure of a data table in the storage engine; and A3, setting a dependent manner of the storage engine based on callback functions registered by users; and secondly, enabling the storage engine Key Value SE to be remade into a storage engine supporting multi-process and multi-thread: B1, enabling all global variable to be placed in a thread local storage; B2, replacing an original signal communication mechanism between processes with a communication mechanism between threads; and B3, setting a thread initialization function and an exit function. The method can make maximum use of a mature storage engine of the RDBMS to achieve development of the NoSQL key-value storage engine, avoids a large amount of repeated development work, and simultaneously can rapidly achieve production of the NoSQL key-value storage engine with high concurrency, complete ACID transactional semantics and an efficient data storage and access method.

This invention discloses a method and a device for realizing high-available distributed queue service. The method comprises the following steps: when a queue interface server receives data sent by a client, requesting Zookeeper server cluster to create persistent sequence nodes of the data; after the nodes are created, storing the data into a local memory queue; when a request for acquiring the data sent by the client is received, extracting queue head data from the local memory queue and requesting the Zookeeper server cluster to delete the persistent sequence nodes of the data; after the nodes are deleted successfully, feeding back the extracted data to the client; and when a node change in the Zookeeper server cluster is monitored, synchronizing a data change caused by the node change into the local memory queue. This invention provides a high available, high extensible and high subsequent queue service realizing plan, and provides a reliable choice for the enterprise application integration and the distributed business collaboration.

The invention provides a high-concurrency transaction request processing method, system and device and a storage medium, and the method comprises the steps: receiving a transaction request, and obtaining commodity identification information; judging whether a commodity to which the transaction request points exists in a preset hot commodity list or not; and if so, distributing the transaction request to a first business processing cluster, deducting the commodity inventory number in the first inventory database by the first business processing cluster, otherwise, distributing the transaction request to a second business processing cluster, and deducting the commodity database in the second inventory database by the second business processing cluster. By adopting the scheme provided by theinvention, the commodity types pointed by the transaction requests are distinguished according to the hot commodity lists, the hot commodities and the common commodities are processed in different server clusters, and the inventory is managed by different inventory databases respectively, so that a high-concurrency and high-availability transaction request processing scheme is provided.

The invention provides a task allocation method. The method is applied to a task allocation system constructed based on a cloud storage Redis technology. The system comprises a Redis server and a plurality of queues, each queue comprises a standard logic queue and a queue insertion logic queue, and the method comprises the following steps: when a service handling request sent by a client is received, obtaining corresponding client information, service information, network information and queuing information of each queue; Calculating the queuing score of each queue according to a first presetrule, and determining the queue corresponding to the maximum queuing score as a recommended queue; Detecting whether the business handling request conforms to a preset queue inserting condition or not, and distributing the business handling request to a corresponding logic queue of the recommendation queue according to a detection result. The invention further provides a task distribution device and system and a computer readable storage medium. The technical problem that a traditional task distribution system in the prior art is only suitable for small-flow application scenarios can be solved.

The invention discloses a task scheduling method and system for a multi-tenant mode SaaS service cluster environment, belongs to the technical field of SaaS services, and solves the problem that taskscheduling in the prior art cannot meet the requirements of high concurrency and expandability in the multi-tenant mode SaaS service cluster environment. The task scheduling method for a multi-tenantmode SaaS service cluster environment comprises the following steps: centralizing tasks in used application examples in a task database; obtaining tenant IDs to complete packaging of the dynamic taskinformation; obtaining tenant IDs, obtaining application instances from the application instance cluster list, calling static tasks by taking the tenant IDs as entry parameters to create APIs, and packaging the static tasks by the APIs; performing clock synchronization of cluster node task scheduling through a database, and in a multi-tenant cluster environment, using a Quartz assembly for task scheduling execution. High-concurrency and extensible task scheduling is realized in a multi-tenant mode SaaS service cluster environment, and the efficiency of task scheduling and task processing is improved.

The invention discloses a management method for a virtual private network of multiple distributed routers. The method specifically comprises the steps that S1, a service processing module collects VPN node data, wherein the node data is the data of VPN nodes for connection of the distributed routers and a VPN core processing module; S2, the service processing module stores the VPN node data in a database; S3, the VPN node data is visually displayed through an HTTP service module; and S4, the service processing module modifies the VPN node data according to a request of a client. The invention aims at providing the management method for the virtual private networks of the multiple distributed routers. According to the method, the VPN virtual network can be visually displayed and networked, and the VPN virtual network can be professionally maintained by a user. Moreover, the invention also discloses a system corresponding to the method.

The invention relates to a mobile-end IM signal synchronization method based on the global monotonic serial number. The synchronization method is applied to an enterprise IM system which is composed of mobile equipment clients and a server, the server comprises a connection server, a message processing center and a memory, and each mobile equipment client corresponds to one group of members. The method can be used to synchronize IM messages among group members in the enterprise IM system, and comprises the steps that one group member establishes a personal monotonic serial number queue, the group member transmits an IM message, and other group members receive the IM message according to corresponding personal monotonic sequence number queues. Compared with the prior art, the synchronization method has the advantages including high availability and high response speed.

The invention discloses a lightweight message middleware system of a virtual test range verification system. The lightweight message middleware system comprises a middleware message forwarding module,a middleware model service module, a middleware time synchronization service module and an object model transceiving module, the middleware system solves the problems that existing virtual test rangetest equipment is difficult to interact, slow in information transmission, low in message concurrency and the like. The invention further provides a virtual test range system communication method based on the message-oriented middleware system. The method specifically comprises the following steps: S1, creating an object model component which can be stored and called by middleware; S2, uploadingthe middleware object model component of the virtual test target range system to a middleware storage server; S3, performing permission and other information binding on the object model component; S4,activating the object model component before the middleware of the virtual test target range system starts; S5, synchronizing the time when the equipment needs to be interoperable; and S6, performinginformation interaction of the virtual device.

The embodiment of the invention provides a data processing method and device, and the method comprises the steps of completely reading to-be-processed business data from a target database, and storingthe business data into a plurality of caches according to a preset classification rule; creating a plurality of data processing tasks according to the classification result of the business data; andallocating the plurality of data processing tasks to a preset number of data processing threads to trigger the data processing threads to process the service data in the corresponding caches based onthe data processing tasks. The method comprises reading to-be-processed business data from a database in batches in advance; loading all the read data into a plurality of caches for classification andarrangement; in the program running process, taking the to-be-processed service data out of the caches to be processed through the multiple processing threads, so the multiple processing threads executed in parallel are directly in butt joint with the multiple caches, and therefore pressure on a database cannot be caused by increase of the number of the processing threads, high concurrency and high expansion are achieved, and the running efficiency of data processing is greatly improved.