9results about How to "Reduce request latency" patented technology

This invention discloses an application creation method, which includes the following steps: when creating a target application using BestEffort Pod, obtaining the initial number of CPU cores to be allocated to the target application; obtaining a preset number of CPU NUMA affinity initialization cores; performing CPU NUMA affinity operation based on the number of CPU NUMA affinity initialization cores to obtain an affinity result; when it is determined that the CPU core binding condition is met based on the initial number of CPU cores and the affinity result, determining the initial CPU cores to be bound; and binding the target application to the initial CPU cores to create the target application using the initial CPU cores. Applying the application creation method provided by this invention improves the ability of BestEffort Pod to efficiently execute business logic, reduces request latency, and improves application running efficiency. This invention also discloses an application creation device, equipment, and storage medium, which have corresponding technical effects.

The invention provides a data access method and device, terminal equipment and a storage medium, and is suitable for the technical field of data processing, and the method comprises the following steps: obtaining data access request information; determining account period information of data to be accessed according to the data access request information; judging whether the account period information of the to-be-accessed data is within a preset account period range or not; if yes, taking the memory database as a data access range; if not, taking the basic database as a data access range; and accessing to-be-accessed data according to the data access request information and the data access range. According to the method and the device, under the condition of limited software platform construction cost, the memory database and the basic database are used for separating the data of different account periods, so that the data access efficiency is improved, and mutual interference among different databases is avoided.

The present invention discloses a collaborative caching method based on quantized federated reinforcement learning in a fog wireless access network, including: 1. Initializing the local cache content of the cache node according to the global content popularity, initializing the local model weight parameter of the cache node, the number of local model iterations in a single cycle and the total model training cycle; 2. Each cache node makes content cache decisions and local model training according to the received user request information; 3. Calculate the cache hit rate and user request delay; 4. At the end of a training cycle, perform network pruning on the local model of each cache node; 5. Perform quantized compression and weight sharing processing on the local model update amount; 6. Upload the local model update amount to the cloud center for global model aggregation to obtain a global model, and then distribute the global model to each cache node to enter the next training cycle. The present invention improves the cache hit rate and reduces the network load pressure.

The invention belongs to the technical field of service scheduling, and discloses a service scheduling method and device, a scheduling server and a storage medium. The method comprises the following steps: when a target questioning request sent by a client is received, obtaining a current load state of each reasoning node of a reasoning server; determining a target node from the reasoning nodes according to the current load state; and scheduling the target question request to the target node, so that the target node responds to the target question request. Through the method, the request time delay of the user can be effectively shortened.

The embodiment of the invention relates to an inference service instance binding method and system, and the method comprises the steps: obtaining instance identification information of at least one candidate inference service instance through an inference client in an inference session establishment process; under the condition that the historical instance identification information is locally stored in the reasoning client, matching the locally stored historical instance identification information with the instance identification information of at least one candidate reasoning service instance; if the matching is successful, determining the successfully matched candidate inference service instance as a target inference service instance, and establishing or maintaining session connection with the target service instance to carry out an inference dialogue; wherein the historical instance identification information is instance identification information of inference service instances connected to the inference client in the last round of inference session; according to the method, when the inference session is newly built, the client can be directly connected to the original instance in which the historical cache is stored, computing resources are reused, meanwhile, a centralized component does not need to be introduced, and the performance bottleneck and the single-point fault risk are avoided.

The application discloses a kind of intelligent DHCP relay system and method based on dynamic network topology perception, system includes: network detection module, configured to send probe packet to multiple DHCP servers, based on the response information of probe packet, obtain the network performance index of each DHCP server between forwarding path;Quality of service evaluation module is configured to receive the network performance index from network detection module, based on network performance index, the comprehensive score of each forwarding path is calculated, and the quality of service map is updated according to comprehensive score, wherein the quality of service map is used to store forwarding path and corresponding quality of service index;Routing decision module is configured to query the updated quality of service map in real time, based on comprehensive score, obtain optimal path, and encapsulate DHCP request message of DHCP client based on optimal path and forward to corresponding DHCP server.The embodiment of the application can reduce request delay through dynamic path selection, improve success rate.

The embodiment of the application provides a long connection task execution method and device and related equipment, and relates to the technical field of Internet of Things. The long connection task execution method can be applied to message pushing, instant messaging service, Internet of Things communication service and the like. The long connection task execution method comprises the following steps: receiving a target request sent by a first client through a long connection channel; the target request carries target task information; calling a target queue thread corresponding to the first client based on the target task information, and executing a target task corresponding to the target request based on pre-stored task information stored by using a dictionary data structure through the target queue thread; based on the target request, feeding back result information of the target task executed by the target queue thread to the first client through the long connection channel; wherein the long connection channel and the first client have a corresponding relationship based on an identity of the client. The embodiment of the application can support high-concurrency requests of the client.

The present invention provides a method and system for jointly optimizing microservice call graph deployment and request routing, the method comprising: obtaining the computing resources required for each microservice, sorting the undeployed microservices according to the required computing resource size, and sequentially allocating the microservices requested by the user to the deployed microservice nodes based on the constraint of the remaining capacity of the server node to obtain a microservice deployment set; clustering and partitioning the tuples of different microservices; sorting the request flow set obtained by clustering in descending order according to the sum of the arrival rates of the requests of each partition, traversing the server nodes of the deployed microservices, allocating the microservice requests in each cluster partition to the server nodes, and directly allocating or splitting the microservice requests to multiple server nodes according to the computing resources owned by the nodes to obtain the microservice request forwarding probability and request routing. The scheme can not only reduce the request service delay and save computing resources, but also ensure the request success rate.

The present invention discloses a service call intelligent management method and system based on a data exchange platform, which relates to the technical field of service call, and comprises the following steps: obtaining a plurality of connection pool sizes to be screened; applying the plurality of connection pool sizes to be screened to intelligent management of service calls, and respectively obtaining corresponding scheduling data and concurrent exception data under high concurrent load; constructing a connection pool size evaluation model based on machine learning according to the scheduling data and concurrent exception data; constructing a scheduling evaluation curve with the connection pool size evaluation model and the corresponding connection pool size; performing data analysis on the scheduling evaluation curve to obtain the screened connection pool size, and applying it to intelligent management of service calls. The present application is used to solve the problem that the connection pool size setting is unbalanced, resulting in waste of system resources or increased request delay, thereby affecting the overall performance and stability of service calls.