71 results about "Request distribution" patented technology

The application relates to a cloud service-based model inference service system and method capable of on-demand scaling. The system comprises a gateway module, a model inference service management module and a request distribution module. The gateway module is used for receiving a user request and extracting heterogeneous features representing the request's demand for computing resources. The model inference service management module is used for generating resource scheduling instructions for each inference service instance based on the multi-dimensional indicators of the deployed inference service instances, in combination with the resource cost and availability information of each cloud resource pool. The request distribution module is used for distributing the user request to the inference service instance that is adapted to the heterogeneous features according to the resource scheduling instructions. The system can solve the problems of low utilization rate of computing resources, poor request adaptability and lack of cross-cloud scheduling in the prior art.

This disclosure provides a request processing method and apparatus, an electronic device, and a storage medium. The request processing method is applied to an artificial intelligence processor and includes: receiving a first request; assigning an identifier to the first request to identify the first request; determining, based on the target storage address of the first request, whether a second request exists in a set including pending requests and requests in execution, wherein the second request is a request having the same target storage address as the first request; in response to the existence of the second request, establishing a dependency relationship between the first request and the second request, wherein the dependency relationship indicates that the first request will be executed after the second request is completed; removing the dependency relationship after the second request is executed; and executing the first request. This method can establish dependencies between requests, enabling the serialized execution of requests.

The application discloses a power industrial control safety protection system and method based on a microkernel active defense, wherein the system takes a microkernel isolation base as a minimum trusted computing base, allocates revocable communication endpoints and controlled mapping permissions to each protection domain, and provides base support for the isolation domain boundary and permission recovery in the active defense; in the method, the system is uniformly formatted by an input module to process field data and service requests and generate a request identifier, a request distribution service distributes the same request to an odd number of online executors for parallel processing, a single decision output available externally is generated, an executor scheduling service selects a candidate executor from a candidate protection domain pool and reconstructs an online user mode service cluster according to a feedback index by adopting a periodic rotation and event triggering strategy, and if the request distribution service, the consistency arbitration service or the executor scheduling service fails, a minimum bottom-up cleaning is performed by an active defense kernel state support module and a reestablishable state is entered.

This invention discloses an AI intelligent system with multiple output data that can reduce erroneous selections, comprising: a registration module for inputting a device type or payment type registered by a user; an initialization module connected to the registration module, which configures multiple different artificial intelligence models according to the device type or payment type, wherein each artificial intelligence model is based on a large language model; an input module connected to the initialization module for inputting a user request; a processing module connected to the input module, which assigns the user request to the multiple different artificial intelligence models; a result generation module connected to the processing module, which generates multiple results from the user request using the multiple different artificial intelligence models; and a result display module connected to the result generation module, which displays the multiple results.

This invention relates to the field of industrial communication technology and discloses a communication method and apparatus for industrial switches to connect walkie-talkies without a gateway. The method involves a management unit receiving a walkie-talkie terminal registration request, assigning a unique virtual LAN identifier, configuring a multicast group address, and establishing a binding relationship. An analog audio interface module receives analog voice signals and accompanying signaling, performs sampling, analog-to-digital conversion, compression encoding, and parses the PTT trigger signal to generate a transmit enable signal. Data frames are then encapsulated and sent to the switching chip. The switching chip queries the forwarding table based on the transmit enable signal and the walkie-talkie identifier, adds a virtual LAN identifier, and forwards the data frames via multicast. The management unit continuously monitors link quality and automatically switches the channel to a backup port when quality deteriorates, ensuring stable and reliable communication, simplifying the network architecture, and reducing costs.

This application discloses a mobility management method, apparatus, and network-side device, belonging to the field of communication technology. The mobility management method of this application includes: under the condition of satisfying a first condition, a first network entity sends a target message to a second network entity, the target message being used to request the allocation of an Evolved Packet System EPS Bearer Identifier (EBI); wherein the first condition includes: a terminal switching or reselecting from a second mobile communication system to a first mobile communication system; and a target PDU session in the Protocol Data Unit (PDU) session corresponding to the first mobile communication system that can be switched or reselected to a third mobile communication system; wherein the first mobile communication system is a fifth-generation mobile communication system, and the third mobile communication system is a fourth-generation mobile communication system.

The application relates to a resource allocation method and a resource allocation device. The method comprises the following steps: receiving at least one computing task request sent by a power internet of things device; distributing each computing task request to a low-delay network slice and a high-throughput network slice; calculating a first task demand parameter of a first computing task request included in the low-delay network slice; calculating a second task demand parameter of a second computing task request included in the high-throughput network slice; and determining the computing power resource allocated to each computing task request according to the first task demand parameter and the second task demand parameter. The method can improve the utilization rate of the computing power resource.

There is provided a method of computing resource allocation. The method comprises allocating a first bounded amount of computing resources forming a first set of computing resources; exclusively assigning the first set of computing resources to a first process of a computer program; receiving a request from the first process for additional computing resources; in response to the request from the first process, allocating a second bounded amount of computing resources forming a second set of computing resources; and spawning a second process from the first process and exclusively assigning the second set of computing resources to the second process; wherein this method may be repeated indefinitely by the first process, second process, or any other process created according to this method. By following this method, a process does not control the amount of computing resources allocated to that process (i.e., itself), but instead controls the amount of computing resources allocated to its child processes.

This specification provides a request distribution processing method and a request distribution processing system. The request distribution processing method includes: obtaining file processing requests for target files; determining a target request processing queue set from multiple candidate request processing queue sets based on the request type of the file processing request, wherein the multiple candidate request processing queue sets correspond to different request types; determining a target request processing queue from the target request processing queue set based on the file identifier of the target file, and adding the file processing request to the target request processing queue; and invoking the request distribution resource corresponding to the target request processing queue to distribute and process the target requests in the target request processing queue. This eliminates the risk of read requests being starved due to resource contention, ensuring low-latency response for data loading; and ensures that processing requests for massive numbers of small files can obtain a fair scheduling opportunity, thereby preventing small files from being starved by large files.

The present disclosure discloses a data read and write method. The method is applied to a logical management controller, where the logical management controller communicates with a central processing unit (CPU) and at least two expanded memory controllers respectively based on a compute express link (CXL) protocol, the logical management controller includes a request parsing module, an address decoding module, a request decomposition module and a request allocation module, and the method includes: parsing a received data read and write request for target data to obtain a local command; performing command parsing on the local command to determine location information for reading and writing data; generating a plurality of data read and write sub-requests based on the location information; sending the plurality of data read and write sub-requests to target expanded memory controllers; and controlling the target expanded memory controllers to perform data read and write operations in parallel according to the plurality of data read and write sub-requests.

Embodiments of this disclosure relate to methods, apparatus, electronic devices, and media for scheduling accelerator resources. The method includes acquiring an accelerator resource allocation request related to a machine learning model, wherein the accelerator resource allocation request indicates the amount of accelerator resources required by a group of containers running the machine learning model. The method further includes determining a group of containers to be rescheduled according to a rescheduling policy in response to a rescheduling condition triggered by the accelerator resource allocation request. Furthermore, the method includes allocating accelerator resources on a first node for the accelerator resource allocation request by migrating the group of containers to be rescheduled from a first node to a second node. According to embodiments of this disclosure, when an accelerator resource allocation request triggers a rescheduling condition, determining the group of containers to be rescheduled according to a rescheduling policy, and then migrating the group of containers to be rescheduled to another node to satisfy the resource allocation request, can improve the utilization efficiency of accelerator resources.

This application provides an error-aware multi-model scheduling method, apparatus, device, and storage medium. The method includes: acquiring a task request, and the processing success probability and processing consumption parameters of each large model providing object; acquiring the task baseline coefficient of the task request, and calculating a scheduling priority value for each large model providing object based on the task baseline coefficient, processing success probability, and processing consumption parameters; allocating the task request to the target large model providing object with the highest scheduling priority value for processing, acquiring the task processing result for evaluation, and obtaining an evaluation coefficient; determining the reference scheduling priority value and evaluation coefficient of a reference large model providing object with a scheduling priority value lower than the target large model providing object, and calculating the result exchange coefficient of the target large model providing object; and returning resource exchange data to the target large model providing object based on the result exchange coefficient. This improves the accuracy and efficiency of routing resource allocation.

Embodiments of the present disclosure provide a method, device, electronic equipment and storage medium for processing high-concurrency data requests. The method comprises: obtaining a data request sent by a requester; hashing the data request to a first partition in at least one partition; requesting a partition operation lock for the first partition according to the data request; querying partition inventory information of the first partition; in response to no unallocated inventory in the partition inventory, obtaining a batch of inventory from an inventory pool; in response to unallocated inventory in the partition inventory, allocating the unallocated inventory to the data request; and feeding back a data request result to the requester. The method allocates the inventory in the inventory pool to each partition in batches to achieve dynamic balancing of inventory allocation among each partition according to the inventory allocation speed of each partition, thereby achieving fast and balanced inventory allocation. At the same time, the overall inventory allocation complexity is reduced, and the overall work stability is enhanced.

This application relates to the field of wireless communication technologies, and discloses a method for transmitting a sidelink positioning reference signal, a terminal, and a network side device. The method for transmitting a sidelink positioning reference signal in embodiments of this application includes: A terminal determines that an SL PRS needs to be sent. The terminal sends a target request to an access network device, where the target request is used to request allocation of a sidelink resource, and the sidelink resource is used to send the SL PRS. The terminal obtains a sidelink resource allocated to the terminal by the access network device. The terminal sends the SL PRS on the allocated sidelink resource.

The application relates to the field of communication technology and discloses a programmable switch user call communication method and system, which comprises the following steps: acquiring an actual processing time interval of a digital signal processor in a programmable switch for voice data packets, determining a processing deviation of the digital signal processor based on the actual processing time interval, analyzing the processing deviation to determine whether the digital signal processor is in a performance decline mode, adjusting a routing priority of the digital signal processor if it is determined that the digital signal processor is in the performance decline mode, receiving a new call request, and allocating a digital signal processor for the new call request according to the real-time routing priority of the digital signal processor. The application combines real-time performance monitoring and dynamic resource scheduling, so that the whole system can better adapt to complex and changeable operation environments and improve the reliability of long-term operation.

The specification provides a load balancing-based request distribution method and device, and relates to the technical field of load balancing. The method comprises the following steps: monitoring the connection quantity of a plurality of service instances, the service instances corresponding to weights; determining whether the plurality of service instances satisfy a balance condition according to the connection quantity, the balance condition comprising: the proportion of the connection quantity between the service instances being equal to the proportion of the weights between the service instances; if the balance condition is not satisfied, selecting an abnormal service instance from the plurality of service instances; adjusting the weight of the abnormal service instance, so as to distribute connection requests to the abnormal service instance according to the adjusted weight, and make the connection quantity of the plurality of service instances satisfy the balance condition. The above method can avoid the problem of load inclination caused by uneven request distribution.

The embodiment of the present application provides a kind of service management method and device, and the order page of first service business is shown by the graphical user interface of first terminal, and order page includes at least one service control, each service control corresponds to different categories of service business, the method includes: receiving the first operation of target service control in at least one service control;According to first operation, the second service business associated with target service control is acquired, and the combined service request corresponding to first service business and the second service business associated with at least one target service control is generated;Combined service request is sent to server, and the notification information fed back by server after successfully matching target receiver is received and shown.The present application can be realized in the same order page, and different service businesses with relevance are added, and combined service request is sent to server, and combined service request is distributed to the same merchant terminal by server once, and distribution resource is saved.

The operation of the elevator will continue without interruption, even when the control entity for the group management of the elevator is switched. [Solution] The backup system comprises a second group management control device that can replace the first group management control device, and a transmission processing unit. The second group management control device is assigned a different IP address than the first group management control device. The transmission processing unit sends a request to assign a platform call for that user to either the first group management control device or the second group management control device each time a user registers a destination floor at the destination floor registration device. When the first group management control device is operating normally, the group management function of the second group management control device, which assigns platform calls in response to requests from the destination floor registration device, is set to "disabled". Then, when the first group management control device becomes unable to operate normally, the second group management control device switches its own group management function setting from "disabled" to "enabled".

The application provides a read-write separation analog signal control method and system, comprising: separating read analog signal quantity and write analog signal quantity to different asynchronous message buffer queues, associating multiple concurrent read or write processes with the different asynchronous message buffer queues, and achieving mutual non-interference of the read analog signal quantity and the write analog signal quantity through configurable read-write load balancing algorithm design, distributing a large number of parallel read-write requests to different read-write processes through optional load balancing algorithm, ensuring timeliness of the write analog signal quantity, and realizing read-write separation of the analog signal quantity. The application solves the problem that analog signal quantity can be timely written to equipment under the condition of a large number of concurrent read-write requests in industrial internet; mutual non-interference of the read analog signal quantity and the write analog signal quantity is achieved through configurable read-write load balancing algorithm design; and timeliness of the write analog signal quantity is ensured.

A method (400) and system (100) for analyzing traffic in a communication network (110) is disclosed. The method (400) comprises receiving a mirrored traffic data of the traffic in the communication network (110). The method may comprise adding metadata to each of the plurality of requests to sequentially log the traffic data. The method (400) may further comprise identifying and transforming sensitive information in each of the plurality of requests to maintain privacy of the traffic data. The method (400) may further comprise assigning a unique Identifier (ID) to each of the plurality of requests. Further, the method (400) further comprises analysing the plurality of requests to identify one or more patterns in the traffic data. The method (400) comprises assigning a pattern Identifier (ID) to each of the one or more patterns. Further, the method may comprise storing the plurality of requests and associated information in a database.

Embodiments of the application disclose a transaction processing method, device, equipment and medium. The method comprises: extracting a global transaction number from an obtained sub-transaction request, and allocating a sub-transaction number to the sub-transaction request; writing an association between the global transaction number and the sub-transaction number into a transaction association table; processing the sub-transaction to obtain a sub-transaction processing result, and persisting the sub-transaction processing result and controlling the sub-transaction processing result to be in an invisible state to the outside; in response to a commit instruction of the global transaction, determining at least two sub-transaction numbers associated according to the transaction association table, and switching the sub-transaction processing result corresponding to the at least two sub-transaction numbers to a visible state to the outside. The scheme can synchronize data in parallel, has superior concurrent performance, improves synchronization efficiency, and guarantees consistency of data read by a downstream database at any time. Meanwhile, the internal processing mechanism of the scheme is simple, and resource consumption is small.

We propose a collaborative logistics system suitable for enabling logistics to be realized through cooperation among businesses belonging to different logistics economic zones. [Solution] The collaborative logistics system 1 enables logistics through the collaboration of multiple logistics economic zones. The collaborative planning processing unit 17 on server 3 creates a collaborative operation plan by assigning resources responsible for transportation based on requests in one logistics economic zone to also handle transportation based on requests in a different logistics economic zone. The confidentiality management unit 18 on server 3 distributes the portion of the collaborative operation plan and resource identification data related to the request to the business device corresponding to the logistics economic zone that received the request, distributes the portion related to the transportation that the business device is responsible for to the business device corresponding to the logistics economic zone that did not receive the request but is responsible for transportation, and does not distribute the portion related to the request to the business device corresponding to the logistics economic zone that did not receive the request and is not responsible for transportation.

The application discloses a load balancing method and device, medium and product. The method comprises the following steps: determining the index data of a service node based on a preset collection strategy, evaluating the index data, determining the node state type and comprehensive score of the service node; in response to a batch of user requests, determining the target proportion of request distribution according to the node state type and comprehensive score of the service node, and determining the matching result of the batch of user requests and the service node; forwarding the batch of user requests to the corresponding service node according to the target proportion and the matching result, so that the corresponding service node processes the received user requests, and load balancing is realized. The technical scheme of the application can comprehensively evaluate the service capacity of the node, accurately match the requests and the node according to the evaluation result, improve the business processing efficiency, and realize effective load balancing.

The present disclosure discloses a batch computing system and an associated method. The batch computing system includes a memory, a task manager and an inference computer. The memory stores a shared model parameter set common to a plurality of tasks that is generated by fine tuning a shared model and a task-specific parameter set of each task. The inference computer receives a plurality of task requests, derives a data length and a designated task of each task request, and enables the task manager to read a task-specific parameter set and a shared model parameter set corresponding to each task request. The inference computer further assigns task requests corresponding to the shared model to a plurality of computation batches, performs, in batch, the common computation of designated tasks in each batch computation, and performs task-specific computation operations of the designated tasks of each batch computation.

Embodiments of the present application provide a data query method and device, a computer readable storage medium and an electronic device. The method comprises: determining a target data model corresponding to a current query request; determining a target data source according to a service priority of different available data sources in the target data model under the current query request; and outputting a query result according to the current query request through the target data source. The technical solution of the embodiments of the present application can determine a target data source for executing a query request according to a service priority of different available data sources in a target data model under the current query request, thereby reasonably allocating data sources for the query request and improving the query experience of users.

This disclosure relates to wireless radio link recovery methods. In various implementations, the method includes determining a beam failure event for a first cell, and transmitting a beam recovery request message on a shared uplink channel when available. In other implementations, the method includes determining that the shared uplink channel is not available for transmitting the beam recovery request message and transmitting a resource request to request allocation of the shared uplink channel via a second cell.

The application provides a buffer capacity processing method and device, electronic equipment, medium and product. It relates to the field of distributed storage. The method comprises the following steps: determining the request type of a network request, the request type comprising an internal request and an external request; according to the request type, adopting a preset corresponding buffer allocation strategy to allocate an initial buffer for the network request; in the data receiving process, detecting the usage state of the initial buffer; if it is detected that the capacity of the initial buffer is insufficient, dynamically adjusting the capacity of the initial buffer. Based on the buffer capacity processing method provided by the application, the buffer size can be dynamically adjusted, thereby significantly improving the performance and reliability of the distributed storage system.

This application discloses an inference service request scheduling system, method, and electronic device, relating to the field of artificial intelligence technology. Because the routing module can score and preferentially schedule the decoding service instance cluster, replacing the traditional round-robin mechanism, it considers the load and inference resources of each instance during request distribution. Simultaneously, due to the addition of a prefix coverage condition judgment mechanism, if it is determined that the current key-value cache of the target decoding service instance with the highest first scheduling score is sufficient to complete the original inference request, the original inference request is directly sent to the target decoding service instance. This not only improves the cache reuse rate of the decoding service instance but also reduces unnecessary redundant calculations and data transmission, thereby improving the response efficiency of inference requests and reducing the response latency of inference requests.

This specification discloses an error request scheduling method and system for distributed applications. The method includes a load balancing layer responding to a client's request by distributing the request to a first gateway node in the gateway layer. If an error matching a first set of fault types occurs during request distribution at the first gateway node, the request is sequentially distributed to a second gateway node in the gateway layer until a successful request distribution occurs at a target second gateway node. The target second gateway node in the gateway layer sends the request to a first microservice instance in the microservice layer. If an error matching a second set of fault types occurs during a call to the first microservice instance, the request is sequentially sent to a second microservice instance in the microservice layer until a successful call occurs at a target second microservice instance. In this embodiment, a multi-level collaborative fault-tolerance system is constructed, enabling proactive fault-tolerant scheduling of error requests, minimizing the unavailability of the business system and improving the overall user experience.