Patents
Literature
Patsnap Eureka AI that helps you search prior art, draft patents, and assess FTO risks, powered by patent and scientific literature data.

49 results about "Tertiary storage" patented technology

Tertiary storage or tertiary memory is a level below secondary storage. Typically, it involves a robotic mechanism which will mount (insert) and dismount removable mass storage media into a storage device according to the system's demands; such data are often copied to secondary storage before use.

Methods and multi-site systems to provide recovery point objective (RPO) protection and automatically initiate realignment and reconfiguration of a protection configuration from the secondary storage site to the tertiary storage site upon primary storage site failure

A computer-implemented method comprises providing a synchronous replication relationship from one or more storage objects of a first storage node to one or more replicated storage objects of a second storage node, providing an asynchronous replication relationship with an asynchronous update schedule from the one or more storage objects of the first storage node to one or more replicated storage objects of the third storage node to provide a protection configuration, tracking, with the third storage node of the tertiary site, a state of the secondary storage site, automatically performing a failover from the primary storage site to the secondary storage site and activating a synchronous mirror copy for the one or more replicated storage objects of the second storage node, and automatically initiating realignment and reconfiguration of the protection configuration to the tertiary storage site based upon the state of the secondary storage site.
Owner:NETAPP INC

Database hot loading method and device based on persistent memory, equipment and medium

The invention relates to the technical field of computers, in particular to a database hot loading method, device and equipment based on a persistent memory and a medium, and the method comprises the following steps: constructing a three-level storage structure comprising a dynamic random access storage layer, a persistent memory layer and a solid state disk storage layer; when the current database is in a starting stage, loading the pre-write log from the persistent memory layer, and starting to execute a redo operation and / or a cancel operation from a preset check point so as to recover the state of the current database; and when the current database is in a running stage, obtaining a current data access frequency and a current resource state of each layer in the three-level storage structure, and carrying out cross-layer migration on data in the three-level storage structure based on a preset dynamic scheduling strategy according to the current data access frequency and the current resource state. Therefore, the problems that a traditional database system is long in starting time, high in memory cost, large in resource occupation and the like are solved, and the overall performance and the resource utilization efficiency of the database system are improved.
Owner:TSINGHUA UNIVERSITY +1

Methods and multi-site systems to provide recovery point objective (RPO) protection, snapshot retention between secondary storage site and tertiary storage site, and automatically initiating realignment and reconfiguration of a protection configuration from the secondary storage site to the tertiary storage site upon primary storage site failure

Multi-site distributed storage systems and computer-implemented methods are described for providing common snapshot retention and automatic fanout reconfiguration for an asynchronous leg after a failure event that causes a failover from a primary storage site to a secondary storage site. A computer-implemented method comprises providing an asynchronous replication relationship with an asynchronous update schedule from one or more storage objects of the first storage node to one or more replicated storage objects of the third storage node, creating a snapshot copy of the one or more storage objects of the first storage node, transferring the snapshot copy to the third storage node based on an asynchronous mirror policy, and intercepting the snapshot create operation on the primary storage site and transferring the snapshot copy to the second storage node to provide a common snapshot between the second storage node and the third storage node.
Owner:NETAPP INC

Intelligent knowledge base system based on multi-domain model collaboration and construction method

The invention discloses an intelligent knowledge base system based on multi-domain model collaboration and a construction method, and relates to the field of artificial intelligence knowledge management. The system realizes accurate generation and dynamic management of cross-domain knowledge through a three-level driving architecture of an intelligent scheduling layer, a knowledge production layer and an intelligent knowledge base. The intelligent scheduling layer schedules professional field models such as medical treatment (such as Med-PaLM 2) and law (such as a power law intelligent contract review model) at a millisecond level through a field intention analysis engine and a three-dimensional matching matrix; the knowledge production layer completes format standardization, conflict verification and timeliness marking by using a knowledge cleaning assembly line; the intelligent knowledge base adopts a three-level storage system of a core library (high frequency and high reliability), an expansion library (medium confidence) and a temporary library (to be verified), and combines a dynamic weight algorithm of dominant score (60%), recessive behavior score (30%) and domain expert calibration (10%) to realize real-time optimization and upgrading and downgrading of knowledge.
Owner:宋梓铭

Virtual robot remote interaction method and system based on digital avatar

The invention relates to a remote interaction method and system for a virtual robot based on a digital avatar, and the method comprises the steps: constructing a partial virtualization data set, and enabling the partial virtualization data set to meet a construction condition; and constructing virtualization data by using each part of virtualization data in the part of virtualization data group meeting the construction condition, acting the constructed virtualization data on a virtual robot model, and performing interaction and / or display by using the virtualization data by a virtual robot. According to the method, support is provided for quick response of the virtual robot based on the digital avatar through behavior data splitting and three-level storage resource management, and the method is suitable for quick response during collaborative or parallel operation of multiple virtual robots in a large-scale virtual scene.
Owner:HANGZHOU YUANMEI TECH CO LTD

Computer data writing system and method

The invention relates to the technical field of computer data writing, in particular to a computer data writing system and method.The computer data writing system comprises a multi-level storage module, a data receiving and analyzing module, a heat evaluation and layering module, a differential writing module, a dynamic migration module, a metadata management module and a monitoring and scheduling module; according to the corresponding method, a hot, warm and cold three-level storage architecture is constructed, an initial data storage hierarchy is determined based on a hot degree prediction model, and differential write-in operation is executed; establishing a'migration forecast-write optimization 'cooperation mechanism, a'write performance data-storage hierarchy parameter' closed-loop optimization mechanism and a'multi-scene write strategy-hierarchy-migration 'parameter mapping system to ensure data access transparency; the cold data storage cost is reduced while the high write-in performance of the hot data is ensured, the problem that the storage performance and the cost are difficult to balance in a time sequence big data scene is solved, and the method is suitable for high-frequency time sequence data storage scenes such as power monitoring and financial transactions.
Owner:CHONGQING COLLEGE OF ELECTRONICS ENG

Local data security storage method for non-networked vehicles

The invention discloses a local data security storage method for a non-networked vehicle, which comprises the following steps of: setting a vehicle-mounted MP5 (Mobile Pentium 5), and storing local data through the vehicle-mounted MP5; setting a three-level storage architecture for the vehicle-mounted MP5, and storing vehicle static information into a configuration data area; storing the operation parameters into a runtime data area according to a time sequence; storing the fault event into an event data area; implementing a hierarchical encryption strategy on data, performing full disk encryption on storage equipment, performing independent file encryption on sensitive data files, and implementing additional character encryption on key fields; an encryption key is derived based on a vehicle VIN code and a user PIN, a circular buffer management mechanism is established, and a secure data export interface is provided. According to the invention, the confidentiality, integrity and availability of vehicle local data can be improved, and the safety and reliability of vehicle data storage are improved.
Owner:ANHUI JIANGHUAI AUTOMOBILE GRP CORP LTD

Data conversion methods, apparatus and computer-readable storage media

A data conversion method, apparatus, and computer-readable storage medium are disclosed. The data conversion apparatus includes at least one first-level memory, at least one second-level memory, and at least one third-level memory, which are sequentially connected. The method includes: receiving a first input, the first input including a number of input channels and / or a number of output channels; selecting a corresponding second-level memory according to the number of input channels and / or the number of output channels, and controlling a first read signal and a second read signal according to the number of input channels and / or the number of output channels; storing data of the input channels in the at least one first-level memory; when the first read signal is enabled, reading data from the first-level memory and writing it to the corresponding second-level memory; and when the second read signal is enabled, reading data from the corresponding second-level memory and writing it to the third-level memory.
Owner:BOE TECHNOLOGY GROUP CO LTD +1

Data query method, device, computer equipment, readable storage medium and program product

The present application relates to a data query method, device, computer equipment, computer-readable storage medium and computer program product. The method comprises: querying the product introduction from the first-level storage space based on the target customer identification and target product identification of the product introduction query request; the first-level storage space includes the transaction information of the product that generated transactions within the target time period and the customer during the last transaction within the target time period; when the query fails, querying the product introduction from the second-level storage space based on the target customer identification and the target product identification; the second-level storage space includes the transaction information of each product during the last transaction with the customer; when the query fails again, querying the product introduction from the third-level storage space based on the target customer identification and the target product identification; the third-level storage space includes the transaction information of each product during each transaction with the customer; and returning the product introduction to the terminal. The use of this method can improve the efficiency of data query.
Owner:KINGDEE DEEKING CLOUDCOMPUTING CO LTD

Three-level storage method suitable for industrial control data

The invention discloses a three-level storage method suitable for industrial control data, which is based on cache storage, real-time storage, historical storage, cache monitoring object service and data dump object service, and comprises the following steps of: acquiring latest industrial control data from each data measurement point by an industrial control system data acquisition module; pushing the collected latest industrial control data into a system cache storage in real time according to a cache storage rule, and updating and identifying data measuring points with changed industrial control data; the cache monitoring object service scans the industrial control data of all the data measuring points in real time, reads the changed industrial control data, stores the changed industrial control data into a real-time database in batches in real time, and identifies the data measuring points as updated; and the data dump object service reads the industrial control data conforming to the rule in the real-time database according to the configured dump rule, stores the industrial control data in the historical database according to the dump rule, and deletes the dump industrial control data in the real-time database at the same time to complete the three-level storage processing process of the industrial control data.
Owner:SHANGHAI BAOSTEEL IND TECHNOLOGICAL SERVICE

Attention routing reuse and error-correcting memory systems and page-missing promotion methods

This invention discloses an attention routing reuse and error correction memory system for large language model inference, alleviating the memory wall and computational wall problems caused by the KV cache's GPU memory constraints and the quadratic growth of attention computation. The attention router reuses the attention score calculated during forward propagation as the key-value block relevance score; the sparse scheduling unit performs top-k sparse selection to generate block masks based on this score, driving the sparse attention kernel to skip unselected block computations; the storage scheduling unit migrates key-value blocks to GPU, CPU, and NVMe storage based on the same score; the page fault promotion unit monitors the attention coverage rate, dynamically recalling missed blocks and re-decoding them when it falls below a threshold; the error correction memory layer combines the error prefix and the correction output to form an error correction pair, writes it to the versioned key-value memory layer, and places it at the top for recall, achieving a learning closed loop without modifying model weights. Actual measurements show a 7.284x speedup in 32K context decoding, with coverage increasing from 74% to 92%.
Owner:王皓伊

A multi-event face search system and method based on GPU parallel acceleration and milvus

The application discloses a multi-competition face searching system and method based on GPU parallel acceleration and Milvus, and relates to the technical field of face recognition and vector retrieval. The system is divided into an access layer, a service scheduling layer, a GPU inference engine layer and a three-level storage layer; GPU batch parallel face detection and 512-dimensional feature extraction are realized by relying on CUDA; video frame feature batch deduplication is completed by using high-speed operation of the GPU; Milvus adopts competition_id as a partition key to realize multi-competition data partition isolation, and is matched with Milvus-GPU to accelerate partition vector retrieval; the service scheduling layer dynamically divides GPU computing power priority, and preferentially guarantees real-time retrieval resources; the system configures hierarchical API keys and IP white lists to realize safety management and control, and is compatible with watermark storage and extra_data parameter transparent transmission functions. The application overcomes the defects of low efficiency of traditional CPU serial processing, large video storage redundancy and high full-database retrieval delay, has the advantages of fast processing speed, high hardware utilization rate, small storage cost and strong safety, and is suitable for competition personnel verification, park visitor retrieval, security blacklist control and other scenes.
Owner:JIANGSU CAMBRIAN INFORMATION TECHNOLOGY CO LTD

Optical fiber sensing cardiovascular postoperative wound intelligent monitoring and nursing system

The invention discloses an optical fiber sensing cardiovascular postoperative wound intelligent monitoring and nursing system, and relates to the technical field of medical monitoring and nursing, and the system comprises an optical fiber sensing array module which is customized for different cardiovascular operations and comprises distributed and point type optical fibers; the multi-parameter data acquisition module is used for receiving a signal by using a two-channel demodulator, performing high-frequency data acquisition after interference removal, and synchronizing patient information and vital signs; the wound state analysis module is used for adjusting a threshold value according to the individual condition of the patient and judging the wound state by using an improved support vector machine; the intelligent nursing suggestion module generates a nursing scheme according to the state; according to the user interaction module, the medical care terminal displays data and states, and the patient terminal has voice reminding and AR guidance; and the data storage and update module adopts a three-level storage architecture and periodically updates an algorithm and a nursing scheme library. According to the system, the wound monitoring accuracy and continuity are improved, the medical care management efficiency is optimized, the autonomous nursing ability of a patient is enhanced, and the risk of postoperative wound complications is reduced.
Owner:THE SECOND AFFILIATED HOSPITAL ARMY MEDICAL UNIV

Method and device for managing display card storage resources, electronic equipment and readable medium

The present disclosure provides a graphics card storage resource management method and device, electronic equipment and readable medium, and relates to the technical field of graphics cards. The graphics card storage resource management method comprises: obtaining an access history of a virtual address block; and according to the access history, performing step-by-step migration of the virtual address block between a graphics card, a host memory and a solid state disk in a UVM framework. Through the embodiment of the present disclosure, the solid state disk is organically integrated into the UVM system as the third level of storage, the addressable space of the graphics card is expanded, the access frequency and the characteristics of the storage medium are balanced, and the probability of hot data residing in the high-speed layer is improved. While optimizing the utilization rate of the graphics card storage resource, the graphics card access efficiency and reliability are ensured.
Owner:NANJING TENAFE ELECTRONIC TECHNOLOGY CO LTD

Digital asset storage method, apparatus and device, and storage medium

The invention relates to the technical field of asset data storage, in particular to a digital asset storage method and device, equipment and a storage medium. Generating zero-knowledge proof data according to a preset quantum state superposition principle, the quantum key and the digital assets; performing fragmentation processing on the zero-knowledge proof data according to a preset fragmentation number calculation formula and a preset clustering algorithm; according to the edge storage nodes, the region storage nodes and the cloud storage nodes, constructing a three-level storage architecture; storing the fragmented zero-knowledge proof storage assets to a three-level storage architecture; according to the scheme, quantum keys and zero-knowledge proof are fused, quantum attack is resisted, collusion vulnerabilities are avoided, dynamic fragmentation optimization storage is achieved, differential redundancy is matched, and quantum safety, privacy protection and high-performance storage are achieved through a three-level architecture.
Owner:BAOZHOU DIGITAL TECHNOLOGY (SHENZHEN) CO LTD

An image acquisition system for a drone

ActiveCN224935723UMicrocontrollerControl cell
The utility model discloses a kind of image acquisition systems for unmanned aerial vehicle, it is related to unmanned aerial vehicle data acquisition technical field, solve the problem of existing system storage adaptation deficiency, transmission synchronization difference, high power consumption and low acquisition reliability.The system includes airborne unit and ground base station unit, airborne unit uses STM32 microcontroller module as core, collation image acquisition module, GPS data acquisition module, storage cache module, synchronous control unit, wireless communication module and power management module;Storage cache module uses three-level storage structure, synchronous control unit realizes data transmission synchronization, wireless communication module uses dual-module redundancy design, power management module realizes low-power control, image acquisition module is equipped with anti-shake mounting seat;The utility model improves the collection transmission efficiency, stability and reliability of image and GPS data by hardware structure optimization, reduce power consumption, adapt to the application requirement of small unmanned aerial vehicle, with good practicality and popularization value.
Owner:HENAN HANGRUNMAN INFORMATION TECHNOLOGY CO LTD

FPGA-based efficient parallel autoregressive diffusion model hardware acceleration system

The invention discloses an efficient and parallel autoregressive diffusion model hardware acceleration system based on an FPGA (Field Programmable Gate Array), and relates to the technical field of FPGA and machine learning. The system comprises a host end and an FPGA hardware end, the host end is responsible for preprocessing and post-processing of data, and the FPGA hardware end comprises a system control scheduling layer, a three-level storage structure layer and an autoregressive diffusion parallel architecture. According to the method, a collaborative pipeline design is adopted, deep parallelism of an autoregression Token generation process and a diffusion picture generation process is realized through a Token grouping adding mechanism, and a diffusion denoising process can be started without waiting for the completion of complete sequence generation. Besides, a universal attention processing hardware module is designed in the system, QKV projection and matrix multiplication are realized by using a systolic array, and calculation resources are optimized in cooperation with a pipelined Softmax calculation and delay normalization strategy. Through collaborative design of software and hardware, the problems of high reasoning delay and large resource overhead of the autoregression diffusion model in an edge computing scene are effectively solved, and the throughput and the energy efficiency ratio are remarkably improved while the generation quality is ensured.
Owner:SUN YAT SEN UNIV +1

Heterogeneous data storage and retrieval method and system for electric power operation and maintenance knowledge graph

The invention discloses a heterogeneous data storage and retrieval method and system for an electric power operation inspection knowledge graph, and belongs to the technical field of data retrieval. The method comprises the steps that a three-level storage model of a graph database layer, a column database layer and a time sequence database layer is adopted, and multi-source heterogeneous data are stored in a classified mode in combination with a message queue routing mechanism; a federated query engine is constructed, conversion from a standard SQL to SPARQL and InfluxQL is achieved, and query is optimized through parallel execution and LRU-K cache; a consistent Hash fragmentation strategy based on virtual nodes is adopted, and automatic capacity expansion is achieved in combination with node monitoring. According to the method, the problems of difficulty in heterogeneous data integration, low storage efficiency, poor query real-time performance and limited expansibility in the prior art are solved, efficient storage, low-delay retrieval and flexible expansion of the power operation and maintenance data can be guaranteed, and the method serves for knowledge graph application in the field of power operation and maintenance.
Owner:CHINA ELECTRIC POWER RESEARCH INSTITUTE CO LTD

Calculation method and device for data offline retention time

The invention discloses a data offline retention time calculation method and device, and relates to the technical field of data offline time calculation, and the method comprises the steps: scanning a single physical page of a three-stage storage unit block through return information obtained by decoding a low-density check code in a solid state disk and the threshold voltage distribution characteristic of a flash memory, and obtaining a data offline retention time value of the three-stage storage unit block; the offline time of the solid state disk is estimated, so that the data retention time of the solid state disk and the scanned real-time voltage axis are updated, and the problems that in the related technology, the time cost and the resource overhead of voltage axis scanning on each physical block are large, and due to the fact that the distance between the erasing state and the programming state of the physical block of a single-stage storage unit is long, the time consumption is low are solved. The technical problem that distribution information cannot be obtained in a threshold voltage scanning interval opened by a manufacturer is solved, and the technical effect that the data retention time of the whole disk and the scanned real-time voltage axis are updated in time, so that the accuracy and reliability of subsequent data reading operation are powerfully ensured is achieved.
Owner:INSPUR SUZHOU INTELLIGENT TECH CO LTD

Vector library model heterogeneous scheduling method and system for asymmetric hardware resources

The application discloses a vector library model heterogeneous scheduling method and system facing asymmetric hardware resources, and belongs to the technical field of hardware scheduling. The vector library model heterogeneous scheduling method facing asymmetric hardware resources extracts a computation memory ratio to construct a global asymmetric feature matrix by monitoring physical indexes of nodes in real time, divides the nodes into computation advantage type nodes and capacity advantage type nodes; decouples a vector processing request into reasoning subtasks and retrieval subtasks, and schedules the subtasks to corresponding type nodes based on the global asymmetric feature matrix; for extreme asymmetric nodes, adopts double-buffer asynchronous scheduling based on a three-level storage architecture to completely overlap computation operation and data transfer; monitors resource saturation in real time, and performs subtask cross-node migration and asynchronous compensation computation based on a redirection strategy. The application significantly improves resource utilization and response real-time performance of a heterogeneous cluster under mixed loads of AI reasoning and vector retrieval, reduces transmission delay, and expands model deployment capacity.
Owner:SHANDONG COMP SCI CENTNAT SUPERCOMP CENT IN JINAN

A method and device for storing and managing three-level structures of over-the-counter derivative contract elements

This invention provides a three-level storage management method and apparatus for OTC derivatives contract elements. The method defines a new storage structure for OTC derivatives contract elements, namely, by reclassifying and classifying the contract elements of various types of OTC derivatives contracts according to whether they are common elements and whether they participate in logical control, the method uses a contract model factory to reintegrate the classified contract elements to form a corresponding three-level storage structure. Finally, based on the formed three-level storage structure, the OTC derivatives contract elements are stored and persisted, reducing storage costs while improving R&D efficiency.
Owner:JIANGSU SECURITIES

Data analysis acceleration and optimization processing method based on data service area

The invention relates to a data analysis acceleration and optimization processing method based on a data service area. The method comprises the following steps: step S100, acquiring monitoring data from an acquisition terminal arranged in a data service area; s200, splitting the monitoring data into first-level data, second-level data and third-level data according to a preset data grading relationship; s300, the first-level data, the second-level data and the third-level data are stored in a preset first-level storage space, a preset second storage space and a preset third-level storage space respectively; and S400, performing data integrity verification in the third-level storage space, and deleting the monitoring data after the verification is completed. By means of the method, the accuracy of data storage in the data distributed storage process can be improved.
Owner:STATE GRID ANHUI ELECTRIC POWER CO LTD +1

Data backup management method and system for automobile data recorder

The invention discloses a data backup management method and system for an automobile data recorder, relates to the technical field of data backup management, solves the technical problems of data storage structuralization and safety deficiency and passive and low-efficiency storage space optimization, and aims to calculate the use frequency through a quantitative standard and accurately divide high-frequency / intermediate-frequency / low-frequency scenes. A differential backup strategy is matched, high-frequency scene data backup lagging is avoided, low-frequency scene redundancy is reduced, backup efficiency is improved, a three-level storage system is established, core event data are classified independently, retrieval time is shortened, meanwhile, three-level permission of an administrator, a common user and a temporary authorized user is set, unauthorized access is effectively prevented, and data backup efficiency is improved. The core data privacy security guarantee capability is improved, low-frequency non-important data is transferred to cold storage according to data access frequency and importance dynamic optimization, high-frequency data is encoded by H.265, and the local storage utilization rate is improved.
Owner:SHENZHEN LIUXIN TECH CO LTD

Video data management method, system and storage medium for railway passenger service and production control platform

The present invention provides a video data management method, system and storage medium for a railway passenger service and production control platform, the method comprising: obtaining video code streams and equipment management data of stations under the jurisdiction of the bureau and other bureaus from a comprehensive video monitoring system by a railway bureau-level platform; obtaining video code streams of the station and stations under the jurisdiction of the station from a comprehensive video monitoring system by a station-level platform; generating scene alarm fragment files based on key scenes by the station-level platform, and forming a scene alarm fragment file set, an optimized analysis model set and a relational database on the railway bureau-level platform; and indexing resources based on metadata of the railway bureau-level platform by a national railway group-level platform. The present invention adopts a two-level access and three-level storage application architecture, based on different independent systems to meet the video data access, intelligent video analysis application and data management requirements under bandwidth constraints, and realizes full-line data sharing while avoiding large data volume transmission, while improving safety supervision efficiency and service quality.
Owner:INST OF COMPUTING TECH CHINA ACAD OF RAILWAY SCI +2

Energy consumption use data management system and method based on edge computing

The invention discloses an energy consumption use data management system and method based on edge computing, and relates to the technical field of data management, and the method comprises the steps: collecting energy consumption data through an edge device, binding a geographic coordinate and a timestamp for each piece of data, superposing a device-level feature code, and generating a unique digital fingerprint; when it is detected that the energy consumption value deviates from the self-adaptive threshold value and the feature matching degree is lower than the feature matching degree threshold value, a high-confidence-coefficient exception marking mechanism is automatically triggered, and a hot event record is generated; managing the collected data according to a three-level storage strategy, wherein the three-level storage comprises equipment end storage, edge end storage and cloud end storage; geographically adjacent edge nodes are quickly positioned based on space-time coding, operation states of geographically adjacent equipment are aggregated, a local optimization instruction is generated through a preset strategy rule base, a cloud terminal synthesizes multi-dimensional information to generate an optimization strategy packet, and the optimization strategy packet is directionally issued to a target equipment terminal after virtual verification.
Owner:EXANDS INFORMATION TECH CO LTD

SD-WAN architecture-based wide area network flow management method and system

The invention discloses a wide area network flow management method and system based on an SD-WAN architecture. The method comprises the following steps: constructing a flow table level-to-level management mechanism based on use frequency, and dividing flow table items into a high-priority flow table set and a low-priority flow table set according to use frequency values; constructing a flow table association prediction mechanism, and establishing a prediction model; establishing a multi-stage flow table storage structure at a network node; when the data packet arrives at the network node, flow table entries matched with the data packet are queried in the first-level storage area, the third-level storage area and the second-level storage area in sequence; if the matching is successful, forwarding the data packet by using the matched flow table item, and sending a prediction request to the controller to trigger associated flow table prediction; if the matching fails, requesting to generate a new flow table item from the controller; in order to solve the problem that passive flow table management causes new flow first packet forwarding delay and seriously affects the service quality of delay-sensitive applications, the first packet forwarding delay is reduced to be close to zero delay.
Owner:BEIJING TIANHONG HUITONG NETWORK TECHNOLOGY CO LTD

Data processing method and electronic device

The application provides a data processing method and an electronic device; the method comprises: initializing a memory pool of a first file; wherein the memory pool corresponds to a three-level storage structure, the three-level storage structure comprises a first-level cache ring stored by using a ring queue, a second-level cache unit stored by using a continuous storage space, and a third-level cache slice stored by using a continuous storage space; the first-level cache ring comprises a plurality of second-level cache units; the second-level cache unit comprises a plurality of third-level cache slices; writing data into the memory pool, reading data in the memory pool, or performing remote synchronization on the data in the memory pool. Through the application, the memory utilization and the data access efficiency can be improved.
Owner:CHINA MOBILE (SUZHOU) SOFTWARE TECH CO LTD +1

Data recall information

PendingUS20260186656A1Data fileEngineering
A tiered storage arrangement is contemplated that provides high speed tier 1 storage for rapid access to recalled data, mid-speed tier 2 storage for data that is not frequently recalled and tier 3 storage for archive purposes that may include tape library storage and even cold storage in vaults. Embodiments contemplate a server linked to all tiered storage devices on one end and a host computer on another. Data files can be migrated between the tiered storage systems based on migration policies, such as time elapsed from being filed or last accessed. Also, embodiments contemplate a time bar displayed at the host computer that provides accurate time predictions of data recall including information of the kind of storage on which a target file is stored and the pathway to recall the target file / s locally to the host computer.
Owner:SPECTRA LOGIC CORP

An audio storage method, an electronic device, and a storage medium

PendingCN122337223AComputer hardwareEngineering
This invention provides an audio storage method, electronic device, and storage medium, relating to the field of voice recognition technology. In the method, the audio to be stored is divided into several sub-audio segments, and the emotional intensity value corresponding to each sub-audio segment is determined using its corresponding acoustic and physiological feature vectors. Then, by combining keyword density and topic novelty scores, an importance score is obtained for each sub-audio segment, achieving a comprehensive and accurate quantitative assessment of its importance. Furthermore, a dynamic three-level storage strategy based on importance scores is used to store the sub-audio segments: low, medium, and high importance levels are respectively discarded, compressed, and stored without loss. This achieves refined dynamic configuration of storage resources, eliminating redundant data while ensuring the integrity and availability of the core audio data of the audio to be stored. It significantly reduces storage usage, avoids resource waste, and improves the utilization rate of storage resources.
Owner:QIHUAN TECHNOLOGY (BEIJING) CO LTD

A low-power video linkage recording device for a door lock

The application provides a door lock low-power consumption video linkage recording device, and belongs to the technical field of smart home. The device comprises a door lock main body and a door lock companion. Through the linkage of a heterogeneous control module, a layered wake-up mechanism and a three-level storage unit, and the design of business sharing, the energy consumption of the door lock main body is significantly reduced under the premise of guaranteeing the video recording quality and transmission efficiency. The device has a reasonable structure design, realizes the balance between the door lock endurance and the video monitoring function, and is suitable for various battery-powered smart door lock scenes.
Owner:HANGZHOU XINQIAO INTELLIGENT TECH CO LTD