Patents
Literature
Hiro is an intelligent assistant for R&D personnel, combined with Patent DNA, to facilitate innovative research.
Hiro

35results about How to "Improve lookup rate" patented technology

Load balancing method and system of WEB server cluster

The invention discloses a load balancing method and system of a WEB server cluster. The method comprises the following steps: firstly, separately specifying a to-be-allocated server for each type of static page access, meanwhile constructing a load balancing target function, when a page method request is received, classifying the page method request, and when the page method request is a static page access request, allocating the page method request to a corresponding server according to the type of the page method request; when the page method request is a dynamic page access request, allocating the dynamic page access at various moments within a current period according to an optimum allocation weight of the dynamic page access of the servers within the current period obtained within the previous period; and when the time arrives at the last moment of the current period, obtaining the optimum allocation weight of the dynamic page access of the servers within the next period through the load balancing target function. According to the load balancing method and system disclosed by the invention, different processing modes are respectively provided for static pages and dynamic pages, so the server resources can be fully used, and a load balancing effect can still be achieved under complex access situations.
Owner:SOUTH CHINA UNIV OF TECH

Safety overlay network constructing method of ciphertext full text search system and corresponding full text search method

ActiveCN102855292ASafe and efficient dynamic updateSafe and efficient creationSpecial data processing applicationsFull text searchCiphertext
The invention relates to a safety overlay network constructing method of ciphertext full text search system and a corresponding full text search method. The invention initially provides a safety overlay network concept used for peer-to-peer full test search, solves the problems of distribution, storage and search of massive ciphertext data index files under the premise that recall ratio and precision ratio are not influenced, realizes storage and search of massive data to peer-to-peer nodes of a distributed P2P (peer-to-peer) network and introduces an index file duplicate copying mechanism based on a pivot node, wherein the mechanism ensures that resource can be rapidly obtained by searches in different directions by virtue of a duplicate of the pivot node without increasing occupation of network bandwidth, so that the safety overlay network constructing method and the corresponding full text search method disclosed by the invention are more applicable to peer-to-peer full text search when information amount is large, and efficiency of full text search under the peer-to-peer network environment is further improved.
Owner:GUANGXI UNIV

Trust data management method in peer-to-peer network

The invention relates to a trust data management method in a peer-to-peer network, which provides a safer, available and highly efficient trust data management mechanism for solving the problem of some defects exiting in a traditional data management mode. Differing from past pure centralized type and scattered type management mechanisms, the centralized type and the scattered type are effectively combined by the management mechanism and mutually make up for deficiencies. The equilibrium, the safety and the high efficiency of data access can be achieved by using the trust data management mechanism disclosed by the invention. The method is a method that the centralized type and the scattered type are effectively combined, mutually make up for deficiencies and are improved, and two nodes logically exist in the network. One is a state node which is used for managing a trading state table; and the other is an ordinary node which is a basic element for forming the network and is used for storing a trading record table per se. Trust data management can be safely and efficiently realized by quoting the trading state table and the trading record table.
Owner:NANJING UNIV OF POSTS & TELECOMM

Method and system for improving data search rate in DRAM

The invention discloses a method and system for improving the data search rate in a DRAM, and relates to the technical field of memories. An execution device for data search is provided with a plurality of DRAM devices. The method comprises the following steps: obtaining the total number of the DRAM devices and the number of Bank groups in each DRAM device, wherein each Bank group at least comprises one Bank; performing mirror image copying on the to-be-stored data, and respectively storing the to-be-stored data into each Bank group; and establishing a search rule corresponding to the to-be-stored data, wherein the search rule records the storage addresses of the to-be-stored data and the search sequence corresponding to the storage addresses. According to the invention, the access time sequence limitation of the DRAM equipment is broken through, the data search rate in the DRAM is effectively improved, and the read-write working efficiency of the memory is further effectively improved.
Owner:武汉二进制半导体有限公司

Routing information processing method, device and equipment and readable storage medium

The invention discloses a routing information processing method, device and equipment and a computer readable storage medium. The routing information processing method comprises the following steps: generating a key value pair of each group of routing information; if a first key value pair and a second key value pair exist, wherein the routing node identifiers are the same and the routing group identifiers are different, setting the routing group identifier of the key value pair which is the same as the routing group identifier of the first key value pair and the routing group identifier of a second key value pair group which is the same as the routing group identifier of the second key value pair as the same routing group identifier; otherwise, inverting all the key value pairs to obtain a finally output routing information set. Thus, in the scheme, after each group of routing information is converted into a corresponding key value pair, the key value pairs can be utilized to quickly retrieve the key value pairs with the same routing node identifier and different routing group identifiers, and after the routing group identifiers of the key value pairs are updated, a final required routing information set is obtained through inversion, so that the searching speed is increased, and the searching time is shortened.
Owner:SHENZHEN ONETHING TECH CO LTD

Method and system for classification of network data flow

The invention provides a method for classification of network data flow. The method for the classification of the network data flow comprises building a longest prefix matching (lpm) table, wherein each item in the lpm table is a port range table, and each port range table corresponds to a service; acquiring the destination internet protocol (ip) of service data when receiving the service data, acquiring the destination port of the service data according to the port range tables corresponding to the service data, finding a hash table corresponding to the source ip of the service data, carrying out hash searching according to the hash table to obtain the source ip of the service data, and obtaining the types of the service data according to the source ip of the service data. According to the method for the classification of the network data flow, an lpm algorithm is combined with the hash table, searching speed is high, scalability is good, transmission quality is improved, and transmission cost is reduced. The invention further discloses a system for the classification of the network data flow.
Owner:BEIJING BAIDU NETCOM SCI & TECH CO LTD

Index update method for ciphertext full-text searching system based on dynamic succeed tree index structure

The invention discloses an index update method for a ciphertext full-text searching system based on a dynamic-succeed tree- index structure. The method comprises the followings: add operation, delete operation and modify operation, and the updating granularity is in a file section grade. The adding operation comprises the followings: building leaf information in a relative position for a new added text; deciphering a leaf information set influenced by the added text; inserting new built leaf information into an original index; only altering the related position of the front drive leaf of the added text during the inserting process, so as to enable the position to point to the leaf position of the first character of the added text, and meanwhile writing the original related position value of the front drive leaf into the related position of the last character leaf of the added text; and after each insertion of the new position information, judging the length of the leaf information set, if the length is larger than a preset value, leaf information set division is performed; and encrypting the obtained leaf information. After the adoption of the method, the system is enabled to achieve index creation and dynamic update safely and efficiently under a ciphertext state.
Owner:GUANGXI UNIV

Optimization method and device for searching network server, computer equipment and storage medium

The invention discloses an optimization method for searching a network server based on a consistent hash algorithm, which is applied to the field of network servers and is used for solving the problems of more occupied network resources, high complexity and high implementation difficulty in the existing technology for searching the network server. The method provided by the invention comprises the following steps: receiving a server search request, wherein the server search request comprises a data object to be searched; calculating a hash value of the data object based on a consistent hash algorithm; querying the sum of the current numbers corresponding to all the currently opened network servers from the configuration server, obtaining the numbers of all the currently opened network servers from the sum, and sorting the numbers to obtain a number sequence; inserting the hash value into a sequence ring corresponding to the serial number sequence; and searching a serial number adjacent to the hash value on the sequence ring to determine a corresponding network server.
Owner:HANGZHOU QULIAN TECH CO LTD

Memory isolation method, isolation check circuit and CPU chip

The invention relates to a memory isolation method which comprises the following steps: receiving a memory access request which comprises an access address of a memory page requested to be accessed; searching a table entry corresponding to the access address in a memory paging protection information table, wherein the index value of the table entry corresponding to the access address is determinedaccording to the access address; and performing isolation check on the memory access request based on the access address and the table entry, and when the isolation check succeeds, allowing the memory page to be accessed by the access request. The memory isolation method at least has one of the following technical effects: the address space is divided into the memory pages equal in size in a paging mode, the number of the memory pages can meet the requirement for the number of security virtual machines or security applications of the virtual machines, the index value of the lookup table itemis determined according to the access address in the access request, and the lookup rate of the table item can be remarkably increased.
Owner:HYGON INFORMATION TECH CO LTD

Method, device and equipment for generating active peptide fragment and storage medium

The invention is applicable to the field of biological medicine, and provides a method, device and equipment for generating an active peptide fragment and a storage medium, and the method comprises the following steps: obtaining a peptide fragment generation model corresponding to specific activity, the peptide fragment generation model is obtained by training an LSTM network by using a general peptide data set and a preset active peptide data set corresponding to the specific activity, the universal peptide data set comprises a plurality of universal peptide fragments; and generating a plurality of peptide fragments with specific activity by using the peptide fragment generation model. In the scheme, the peptide fragment generation model is obtained by training the LSTM network by using the general peptide data set and the preset active peptide data set corresponding to the specific activity, and can learn potential characteristics in the general peptide fragment and potential activity rules in the known active peptide fragment in the training process, so that the potential activity of the active peptide fragment is improved. Therefore, in the process of actually using the peptide fragment generation model, the peptide fragments with different specific activities and abundant diversity can be generated according to different specific activity requirements.
Owner:中国科学院深圳理工大学筹 +1

Address table storage and lookup method, device, computer equipment and readable storage medium

The invention provides an address table storage and lookup method, applied to a switch. The method comprises the following steps: receiving an address table query instruction; parsing the address table query instruction to obtain an Mac address and Vlan information to be looked up; querying a mapping address corresponding to the address table query instruction according to the Mac address and the Vlan information; and querying port information corresponding to the address table query instruction in a preset multi-layer chain table according to the mapping address, wherein the port information and the mapping address that corresponds to the port information are stored in the preset multi-layer chain table. The invention also provides an address table storage and lookup device, computer equipment and a readable storage medium. According to the invention, after an address table query instruction is received, the query instruction is parsed to acquire the Mac address and the Vlan information, the corresponding mapping address is queried, and the port information is queried in the preset multi-layer chain table to ensure that the retrieval for the Mac address and the Vlan information can be quickly performed, and the lookup speed can be increased.
Owner:LILING FULLRIVER ELECTRONICS & TECH LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products