40results about How to "Reduce request latency" patented technology

A method is to process packets in a wireless communication system including an evolved core network, a plurality of user devices, and an evolved node B (eNB), the evolved core network including a policy and charging rules function (PCRF), a serving gateway, a caching management gateway and a mobile management entity (MME). The method includes receiving, via the PCRF, the packets; obtaining, via the MME, an user device number; estimating, via the caching management gateway, a connectivity parameter corresponding to the user devices; generating, via the caching management gateway, a decision result according to popularities and sizes of the packets, the user device number and the connectivity parameter; and instructing, via the serving gateway, the eNB and the user devices to cache the packets according to the decision result.

A load-balancer assigns incoming requests to servers at a server farm. An atomic operation assigns both un-encrypted clear-text requests and encrypted requests from a client to the same server at the server farm. An encrypted session is started early by the atomic operation, before encryption is required. The atomic operation is initiated by a special, automatically loaded component on a web page. This component is referenced by code requiring that an encrypted session be used to retrieve the component. Keys and certificates are exchanged between a server and the client to establish the encrypted session. The server generates a secure-sockets-layer (SSL) session ID for the encrypted session. The server also generates a server-assignment cookie that identifies the server at the server farm. The server-assignment cookie is encrypted and sent to the client along with the SSL session ID. The Client decrypts the server-assignment cookie and stores it along with the SSL session ID. The load-balancer stores the SSL session ID along with a server assignment that identifies the server that generated the SSL session ID. When other encrypted requests are generated by the client to the server farm, they include the SSL session ID. The load-balancer uses the SSL session ID to send the requests to the assigned server. When the client sends a non-encrypted clear-text request to the server farm, it includes the decrypted server-assignment cookie. The load balancer parses the clear-text request to find the server-assignment cookie. The load-balancer then sends the request to the assigned server.

A system and method for pre-allocating replicas for a distributed file sharing system. Creating a new file may involve creating a plurality of replicas for the file on a plurality of nodes. In one embodiment nodes in the system may pre-allocate sets of file replicas, where the pre-allocated replicas can be used to satisfy requests to create new files. Pre-allocating the file replicas may decrease the latency of file creation requests by enabling the requests to be satisfied without performing replica allocation in response to the requests.

A mechanism for reprioritizing high-latency input / output operations in a file system is provided. The mechanism expands a file access protocol, such as the direct access file system protocol, by including a hurry up command that adjusts the latency of a given input / output operation. The hurry up command can be employed in the Direct Access File System.

The invention relates to a neighbor cooperation cache replacement method in a content center network. The method is characterized by comprising the steps that step 1, when a node i has a new data einew request, if the cache space of the node i is full is judged; if the cache space of the node i is not full, the step 3 is executed directly; if the cache space of the node i is full, the step 2 is executed, and judging of cooperation cache replacement is carried out; step 2, if the new data einew is cached in a neighbor node of the node i is checked, and if the new data einew is cached, the new data einew is no longer cached in the node i; if the new data einew is not cached, the step 3 is executed; step 3, the new data einew is cached in the local through LRU cache replacement algorithm.

The invention discloses a beam failure recovery request method, which comprises the steps of obtaining channel time-frequency resource allocation information and preamble sequence configuration information for transmission of a beam failure recovery request; selecting candidate downlink transmission beams according to a measurement result; selecting a channel time-frequency resource and / or preamble sequence according to a corresponding relationship between the downlink transmission beams and channel time-frequency resources and / or preamble sequences, the time-frequency resource allocation information and the preamble sequence configuration information; and transmitting the preamble sequence on the channel time-frequency resource. Compared with the prior art, optimization is performed on the resource allocation and process in allusion to the beam failure recovery request, the signaling overhead can be effectively reduced, and the access time delay is reduced.

The disclosed embodiments disclose techniques for optimizing data transfer costs for cloud-based security services. During operation, an intermediary computing device receives a network request from a client located in a remote enterprise location that is sending the network request to a distinct, untrusted remote site (e.g., a site separate from the distinct locations of the remote enterprise, the cloud data center, and the intermediary computing device). The intermediary computing device caches a set of data associated with the network request while forwarding the set of data to the cloud-based security service for analysis. Upon receiving a confirmation from the cloud-based security service that the set of data has been analyzed and is permitted to be transmitted to the specified destination, the intermediary computing device forwards the cached set of data to the specified destination.

A first-in-first-out (FIFO) queue optimized to reduce latency in dequeuing data items from the FIFO. In one implementation, a FIFO queue additionally includes buffers connected to the output of the FIFO queue and bypass logic. The buffers act as the final stages of the FIFO queue. The bypass logic causes input data items to bypass the FIFO and to go straight to the buffers when the buffers are able to receive data items and the FIFO queue is empty. In a second implementation, arbitration logic is coupled to the queue. The arbitration logic controls a multiplexer to output a predetermined number of data items from a number of final stages of the queue. In this second implementation, the arbitration logic gives higher priority to data items in later stages of the queue.

Content-centric networking is one solution of an information-centric networking thought. The invention discloses a centralized caching decision strategy in the content-centric networking. Under the architecture of the content-centric networking, researches of basic functions, such as data naming, content forwarding and intranet caching of information-centric networking, are realized by utilizing the existing framework adopting a SDN; the centralized caching decision strategy is provided; a centralized controller is placed in the content-centric networking; the controller generates one caching list for each node by collecting all node request information, and deterministically guides each node to cache contents; therefore, the transmission flow in the network is minimized; and the user request delay is reduced.

The embodiment of the invention provides a caching method and system. The method includes: for each user in a user group served by any edge node, a request content from a user at a target time is predicted according to a historical request record set of the user within a preset period of time; incomes obtained by caching the request content are obtained and the corresponding request contents are cached to the edge node successively according to the incoming sequence. According to the method and system, with the historical request record from the user, the user request behavior characteristicscan be explored fully and the request content from the user at a certain time in future can be predicted accurately. Moreover, the income obtained by caching each request content is calculated and thecorresponding request content is cached based on the income sequence, so that the cache space of the edge node is utilized efficiently and the request delay is reduced.

The invention discloses a distributed cache method based on mysql binlog. The method guarantees promptness of distributed cache data, reduces pressure of a mysql database, lowers a request delay and improves user experience. The method includes: (1) starting using a binary log file binlog of the mysql database; (2) when data is directly written in the mysql database, writing updated data in the binlog file synchronously; (3) parsing the binlog file, and writing changing data of the database in real time in distributed cache; and (4) when a user accesses, directly acquiring the data from the distributed cache. The invention also discloses a distributed cache system based on mysql binlog.

The invention discloses a method and a system for constructing a consistency hash table based on a non-volatile memory. The method of the invention constructs a consistent hash table, and divides allhash units in the hash table into two layers. The first layer is a hash function accessible unit, the second layer is a hash function inaccessible unit, and the second layer is used for collision processing when a hash conflict occurs in the first layer. The hash cells of each layer are divided into an equal number of n groups, and the groups of the first layer and the second layer with the same number are matched. When the i-th group of the first layer has hash collision, the i-th group of the second layer handles the collision, and the hash table guarantees the consistency of the data by using the 8-byte atomic write technique. The hash table can be restored to a consistent state after a system crash. The invention also realizes a hash table construction system based on non-volatile memory. The hash table constructed by the technical scheme of the invention can reduce the consistency overhead and has low request delay.

The invention discloses a hybrid index-based hybrid memory performance optimization method and system, which belongs to the field of computer data storage, and comprises the following steps of: establishing a hash table in an NVM (Non-Volatile Memory) of a hybrid memory in advance, and establishing a skip table and a waiting queue in a DRAM (Dynamic Random Access Memory); for each operation request, if the operation request is an insertion operation, inserting the operation request into a hash table according to a key value of a key value pair to be inserted, and adding the key value pair andan insertion operation mark into a waiting queue; if the operation request is deletion operation, deleting the key value pair from the hash table according to the key value of the key value pair to bedeleted, and adding the key value pair and a deletion operation mark into a waiting queue; if the operation request is the single-point query operation, obtaining the key value pair from the hash table according to the key value of the key value pair to be queried; and if the operation request is a range query operation, synchronizing all the key value pairs in the waiting queue into the skip list, and obtaining all the key value pairs in the query range from the skip list. According to the invention, various operation requests including range query can be efficiently realized.

The invention relates to a resource management method and system based on edge computing and electronic equipment. The method comprises the following steps: step a, acquiring state information of allavailable edge network resources, and writing the acquired state information of the edge network resources into a resource information database; b, receiving a task request of a client, respectively calculating the task execution time of the current task in the local client and the task execution time of the current task in the edge computing platform according to the related information of the current task and the resource state information of the resource information database, and if the task execution time of the local client is not less than the task execution time of the edge computing platform, executing the step c; and c, matching a target computing node corresponding to the current task through the edge computing platform, executing the current task through the target computing node, and returning an execution result to the client. Resource waste can be effectively reduced, the resource utilization rate is increased, and the task execution time is shortened.

The invention relates to the technical field of communication, and particularly relates to a non-hybrid ant colony routing method in content centric networking. The method comprises the steps of building pheromone tables, wherein each node stores a pheromone table, when a previous ant arrives at a certain node, a port related to a current ant has a corresponding pheromone value, and a port not related to the current ant has no pheromone value; forwarding the previous ant from a source node to an intermediate node, and forwarding the previous ant to other intermediate node by the intermediate node till highest routing hops are achieved or a destination node is found; if the highest routing hops are achieved, executing a pheromone volatile operation; and if the previous ant rinds the destination node, generating a next ant by the destination node, and allowing the next ant to return from the destination node along the forwarding path of the previous ant to the source node and update pheromones of the node passed by. The method keeps the network overhead cost low and prevents premature and stagnation effectively.

An NVMe-oriented IO deterministic optimization strategy method can reduce performance jitter and the like of I / O requests by performing optimization based on sets and window division, thereby improving garbage collection efficiency and further prolonging the service life of a solid state disk, and is characterized by comprising the following steps: step A, realizing mutual independence of different sets by performing set division on NVMe storage units, therefore, reading, writing and garbage collection operations can be performed in parallel among different sets; step B, proposing a new I / O request scheduling algorithm for the NVMe after set division, and avoiding conflict between garbage collection and I / O request access; and step C, designing a new cache management algorithm to perceivethe garbage collection operation of the NVMe set, avoiding the possibility of conflict between the garbage collection operation and the I / O request access to the greatest extent, and reducing the performance jitter of the I / O request.

The invention discloses an intelligent scheduling distributed path calculation method and system, and the method comprises the steps: a detection server initiates a detection data request; the intelligent scheduling platform obtains the heartbeat detection result of the central path calculation server cluster of each region from the monitoring platform, and if the heartbeat detection result of the target central path calculation server cluster of the region where the detection server is located is normal, the target central path calculation server cluster of the region where the detection server is located is abnormal; scheduling the detection data request to the target central path computing server cluster; and a switch in the target central path computing server cluster schedules the detection data request to a central path computing server according to a load balancing mode, so that the detection data is stored through the central path computing server. And synchronizing the detection data to a central path computing server cluster in other regions. According to the technical scheme provided by the invention, the processing efficiency of the detection request can be improved.

The embodiment of the invention discloses an information processing method and device and a storage medium. The method comprises: receiving a private key file issued by a CDN certificate management server d; generating a corresponding private key service based on the private key file; and issuing the private key service to at least one second device, so that the second device provides an encryption or decryption service for a CDN edge node based on the private key service.

The invention discloses a construction method and application of a distributed learning index model, and belongs to the technical field of computer distributed storage. The construction method comprises the following steps: for each storage node, sorting data stored in the storage node according to the size of key values, taking the key values of the stored data as input, taking corresponding sorting positions as output, training a machine learning model to obtain a learning index model of each storage node, and synchronizing the learning index model to all calculation nodes; the calculation node directly modifies data in the storage node through RDMA operation, and a CPU of the storage node does not need to participate in work; meanwhile, the calculation node asynchronously retrains the old model and synchronizes the new model into the storage node; the operation of modifying the data and the model is executed in the calculation node in the distributed system, so that the CPU overhead of the storage node is greatly reduced.

The invention discloses a client lock splitting method based on a distributed file system. The method comprises the steps that resource classification is performed according to an inode number of a file, and metadata caches and additional resources are placed in different resource pools; when request processing operation is performed, the resource pool used in the request processing operation is found according to a module value of the inode number of the file, and the resource pool is locked; and the request processing operation is executed, and the lock of the resource pool is released afterthe operation is completed. Through the method, request response delay is shortened on the basis of improving client IOPS.

The invention discloses a content request method based on fragment cache in an unmanned aerial vehicle ad hoc network. On the premise that each node has the caching capability, the request content is fragmented by using the fragment offset mark in the request message, so that the number of the request messages is reduced, and the data message is acquired from the neighbor node to the greatest extent; therefore, the total delay of data message acquisition is reduced, and the load of a return path is reduced. The method comprises the steps that a destination node sets a slice offset mark in a generated request message; a relay node checks own node cache according to the content in the request message and the slice offset mark, passes back the hit cache slice and modifies the slice offset mark; and finally, the data message corresponding to the chip offset mark is transmitted back to the destination node. In the return process, the relay node can store the cache chips which are not cached. According to the method, a plurality of data messages are obtained by a single request message, so the request delay is reduced, and the data return load is reduced.

The embodiment of the invention discloses a data processing method, a storage medium and electronic equipment. The method comprises the following steps of: receiving a data processing request sent by an API (Application Program Interface) on user equipment through a data receiving end of game host equipment, the data processing request comprising at least one of a data acquisition request and a newly added acquisition index request; controlling the data receiving end to request an SDK to execute the data processing request based on a RPC interface of the SDK; sending a data processing result of the data processing request executed by the SDK to a fixed port of a loop-back network card through a UDP interface; and then controlling the data receiving end to receive the data processing result from the fixed port of the loop-back network card, and sending the data processing result to the API on the user equipment. According to the embodiment of the invention, the data acquisition overhead of the game service can be reduced, the operation of the game process is not blocked, the request delay of players is reduced, and the data acquisition efficiency of the game is improved.