10 results about "Server replication" patented technology

The invention discloses a federal learning model leakage traceability method based on model structure confusion, and the method comprises an initialization step: S1, a server copies an initial model for subsequent structure comparison, and generates a unique binary user identity (UID) for each client; a UID embedding step S2, before each round of training is started, independently executing structure confusion for each client model by the server, embedding a client UID into a model structure, and then issuing the model to the client; a model structure recovery step S3, when one round of training is finished, the server executes structure recovery on each client model, and then executes model aggregation; and S4, when a model ownership owner doubts that a model with an unknown source comes from leakage in a training process, the structure of the model is detected, and a suspicious UID is extracted through structure comparison with an initial model so as to find a leakage person. The method has the characteristics of no negative influence on the performance of the original model, low embedding time overhead, high robustness and the like.

A hot server is identified from a plurality of servers based on one or more server metrics associated with the hot server. A hot data range stored by the hot server is identified based on one or more read density metrics. The hot data range comprises a range of data values with a higher volume of access requests compared to other data values stored by the hot server. The hot data range is replicated across a number of additional servers.

This application provides a method, apparatus, device, and storage medium for transmitting remote messages. It can be used in the field of mobile internet technology. The method is applied to a replication server deployed in a source region, which is communicatively connected to a distribution server deployed in a destination region. The distribution server includes: receiving remote transmission configuration information sent by an operating device; receiving a target remote message sent by a target message cluster in the source region, and encapsulating the target remote message and the remote transmission configuration information to obtain a target message; sending the target message to the distribution server, which then decapsulates the target message to obtain the target remote message and the remote transmission configuration information, and sends the target remote message to the target message cluster in the destination region. The target remote message instructs the target message cluster in the destination region to send the target remote message to a second device.

The application relates to a kind of Nacos cluster lossless upgrade and traffic migration method, system, electronic equipment and computer readable storage medium based on double-center environment.The method comprises the following steps: firstly, the traffic of the first Nacos cluster is copied to the second Nacos cluster through the first Nginx server, and the subscription traffic is proxied to the first Nacos cluster, then the traffic is switched to the second Nginx server through the application load balancer, bidirectional forwarding of the traffic is realized, after the number of instances of the two clusters is stable and consistent, the traffic is completely switched to the second Nacos cluster, and the upgrade is completed.The application discloses a scheme for realizing lossless upgrade and expansion of Nacos cluster, avoiding cluster avalanche, instance decline and instance loss of Nacos cluster, ensuring high availability and stability of Nacos cluster, and realizing no influence on application and transaction during the upgrade process.

The invention provides a server cluster system and a data replication method, which can be applied to the technical field of cluster backup. The server cluster system comprises the steps that a host determines a synchronous replication mode of multiple standby machines based on mirror image attributes and link information of the multiple standby machines; and sending cluster information and the to-be-copied data to a target standby machine in a server copy mode through an optical fiber channel according to a synchronous copy mode of the target standby machine based on a standby machine disk identifier corresponding to the to-be-copied data, respectively sending cluster information and to-be-copied data to a target standby machine in a disk copying mode through a network channel and an optical fiber channel; any standby machine stores the to-be-copied data to a target disk according to the cluster information, and integrates the stored data and the to-be-updated copied data received in the polling time period under the condition that the asynchronous protection mode is determined to be a standby protection mode and the resource occupancy rate is greater than a first preset threshold value; and integrated data is obtained and sent to a backup server through an optical fiber channel.

A hot server is identified from a plurality of servers based on one or more server metrics associated with the hot server. A hot data range stored by the hot server is identified based on one or more read density metrics. The hot data range comprises a range of data values with a higher volume of access requests compared to other data values stored by the hot server. The hot data range is replicated across a number of additional servers.

A method for conserving power and encryption during analysis of media captured by an information capture device using a partitioned neural network includes copying, by an information capture device, an artificial neural network (ANN) from a computer server to the information capture device. Both the ANN on the computer server and the copied ANN include M layers. The method further includes, in response to captured data to be processed being input, partially processing, by the information capture device, the captured data by using the copied ANN to execute the first k layers, where only the k layers are selected to be executed on the information capture device. The method further includes sending, by the information capture device, an output of the kth layer to the computer server, the computer server partially processing the captured data by using the ANN and the output of the kth layer to execute the remaining of the M layers.

Techniques for dynamic application packaging and execution are described herein. In various embodiments, one or more servers, which include one or more processors and non-transitory memory, duplicate a component in an application to generate duplicated components within the application. The server(s) then package the application to include diversified versions of the duplicated components and obtaining metadata describing the diversified versions. The server(s) also receive from a client device a request for execution of a version of the application, compose a unique manifest for the client device in response to the request, where the unique manifest identifies the version of the application and a version of the component among the diversified versions. The server(s) then cause execution of the version of the application at the client device, including causing execution of the version of the component at the client device according to the unique manifest.

This application provides a server cluster system and a data replication method, which can be applied to the field of cluster backup technology. The server cluster system includes: a host determining the synchronous replication mode of multiple standby machines based on the mirror attributes and link information of multiple standby machines; based on the standby machine disk identifier corresponding to the data to be replicated, and according to the synchronous replication mode of the target standby machine, sending cluster information and the data to be replicated to the target standby machine in server replication mode via fiber optic channel, and sending cluster information and the data to be replicated to the target standby machine in disk replication mode via network channel and fiber optic channel respectively; any standby machine storing the data to be replicated to the target disk according to the cluster information; and, if the asynchronous protection mode is determined to be the standby protection mode and the resource utilization rate is greater than a first predetermined threshold, integrating the stored data and the data to be updated and replicated received during the polling period, and sending the integrated data to the backup server via fiber optic channel.

A system of aggregating servers includes a first aggregating server that establishes a session for linking with a first system; a second aggregating server that establishes a session linking with a second system; wherein the first and second aggregating servers replicate mutually structured data provided by the first and / or second system, to provide access to the structured data of the respective systems for the first and second aggregating servers; and coordinate the respective sessions of the system of aggregating servers with the first and / or second system to prevent improper access to the at least first system to be aggregated and / or the at least second system to be aggregated.