42results about How to "Reduce cache pressure" patented technology

The invention provides a method for relieving caching pressure of a server. The method includes the steps that after it is detected that a user logs in, user data recorded in a database are read, and a data model package for the user is constructed and sent to a client so that the client can generate data models; a data maintenance algorithm model is constructed and sent to the client; when data of the user are updated, an algorithm calling command is sent to the client so that the client can call corresponding data maintenance algorithms in the data maintenance algorithm model according to the algorithm calling command, and the data models of the client are updated. The method for relieving caching pressure of the server can effectively solve the problem of pressure maintenance when the server caches user data.

The invention relates to a RPKI (Resource Public Key Infrastructure) data warehouse incremental synchronization method. The method comprises the following steps: 1) an updating element is added in an incremental file, and for an object file of which the content is partially updated, a RPKI relying party performs data synchronization with the RPKI data warehouse through the updating element; 2) an active informing mechanism is added, when the content in the RPKI data warehouse is changed, a new updating informing file is generated and actively sent to the RPKI relying party, thus, the RPKI relying party is enabled to synchronize the updated file; 3) an incremental file monitoring system is established, wherein the system is used for monitoring acquisition condition of the incremental file and deleting the incremental file which is only acquired by a very small amount of the RPKI relying parties. With the method, caching load of the RPKI data warehouse is reduced, and bandwidth pressure during a data synchronization process is relieved.

The present application provides a message scheduling method, device, and network chip, which are applied in a network chip. The network chip includes a classification layer, and the classification layer includes a plurality of classification nodes; and the method includes: when the classification node obtains After the scheduling task of the network message, obtain the node identification of the classification node; according to the mapping relationship between the stored node identification and the queue identification of the output queue, determine the queue identification corresponding to the node identification of the classification node; determine the determined queue Identify whether the corresponding output queue satisfies the backpressure condition; when the backpressure condition is met, the scheduling task is abandoned; when the backpressure condition is not met, the scheduling task of the network message is executed so that the above network message is written to Determine the output queue. By adopting the above method, the dispatching of network messages is realized at the L3 layer, which effectively avoids the congestion problem caused by the excessive buffer size of the output queue, and reduces the buffer pressure of the corresponding output queue.

The invention relates to a method for achieving uniform distribution of all paths of program code streams in the construction process of multiple program TS streams. When TS packages are transmitted at a time, N memory blocks are allocated to N paths of audio / video data packages needing to be transmitted, and list information and the integral number of TS packages are stored in each memory block. The total number of the TS packages in the N memory blocks is counted at the transmitting moment, the percentage of the number of the TS packages in each memory block on the total number of the TS packages is calculated, and a range of consecutive serial numbers is allocated for each memory block according to the corresponding percentage; an uniform distribution random number with the value range smaller than the total number of the TS packages is generated; the TS package, with the generated random falling into the memory block serial number range, in the corresponding memory block is read and transmitted. Uniform distribution random numbers are generated repeatedly until all the TS packages are transmitted. Thus, sudden mass data transmission in a certain path of program is avoided, the stability of broadcasting is improved, and the audio / video data caching pressure at the receiving end of a cable television is relieved.

This application discloses a blockchain business acceptance and business consensus method and device. The consensus node includes a first application, a second application, and a database, and the database is used to store business data processed by the first application and the second application. , in order to reduce the cache pressure of consensus nodes. It can be seen that through the method provided by the embodiment of this application, as long as the consensus node determines the business data corresponding to the business request, it can store the business data in the database and wait to send it to other consensus nodes, so the cache occupation of the consensus node The lower rate increases the number of businesses that consensus nodes can process per unit time, thereby improving the efficiency of consensus verification and processing of business requests by consensus nodes.

The embodiment of the present invention discloses a file uploading method and device. The file uploading method includes: receiving a file uploading request sent by a client, the file uploading request includes complete file content; parsing the file uploading request, and analyzing to obtain Store the data flow fragments in the cache; when the size of the data flow fragments stored in the cache reaches the preset threshold, the micro-thread in the scheduling process writes the data flow fragments in the cache into the storage module, and repeat this step until All data stream segments of the complete file content are written into the storage module. The embodiments of the present invention can reduce cache pressure and improve upload performance.

The invention discloses a method for presenting web AR data based on attribute separation, which belongs to the technical field of computer algorithm processing, and includes the following steps: S10, by separately storing and managing the model and action data at the server end, sending a request at the client end ; S20, the mobile edge computing terminal parses the request, and requests data from the cloud to obtain response resources; S30, performs data calculation on the mobile edge computing terminal, returns the calculation result data to the client, and realizes the dynamic model loading function . The method reduces the pressure of cloud computing and improves performance by adopting web augmented reality call and rendering calculation on the model, providing web augmented reality model service mode in interface mode, and improving the interface operation of mobile edge computing and other technologies.