39 results about "Radix tree" patented technology

Methods, apparatus, and products are disclosed for parallel execution of operations for a partitioned binary radix tree that include: receiving, in a parallel computer, an operational entry for the PBRT, the PBRT comprising a plurality of logical pages that contain a plurality of entries, each logical page included in a tier and containing one or more subentries corresponding to the tier of the logical page containing the subentry, each entry is composed of a subentry from each logical page on an entry path; processing in parallel, on the parallel computer, each logical page in each tier, including: identifying a portion of the operational entry that corresponds to the tier of the logical page, and performing an operation on the logical page in dependence upon the identified portion of the operational entry for the tier; and selecting operation results from the logical pages on the entry path for the operational entry.

A method for processing and verifying remote dynamic data is provided. The method includes providing a radix tree structure having N levels, obtaining and recording N initial values for representing the empty radix tree structure, wherein all nodes at the same level are assigned an identical initial value. When performing a data processing operation to the radix tree structure, determining a first leaf node and calculating and recording the value of each node in a shortest path from the first leaf node to the root node. When performing a verification of a specific data, obtaining a second leaf node corresponding to the specific data, a sibling node of each node in a shortest path from the second leaf node to the root node, and generating a verification result according to a digital signature for verifying the root node, the value of each obtained sibling node, and the specific data.

The invention discloses ART-based (Adaptive Radix Tree based) distributed system graph storage and computing system and method and relates to the technical field of distributed graph computing. The system comprises a data source unit, a data partitioning unit, a data storage unit, and a graph computing unit; the data source unit is provided with a data acquisition module used for acquiring graph data, the data storage unit comprises a database, a file system, a distributed file system and HBase, and the data partitioning unit comprises a data loading module and an adaptive partitioning algorithm module. The system comprises the data source unit, the data partitioning unit, the data storage unit and the graph computing unit; the data source unit is provided with the data acquisition module used for acquiring graph data, the data storage unit comprises the database, the file system, the distributed file system and the HBase, and the data partition unit comprises the data loading module and the adaptive partitioning algorithm module.

A storage server maintains a number of datasets (e.g., exported file systems or other resources). For each dataset, certain clients are allowed to have access (e.g., read access, write access, root access, etc.) and certain other clients are not allowed to have access. Access permission information is maintained to specify which clients are allowed to have access and what kind of access. A method and system are introduced to use a radix tree to store access permission information in a cache, therefore allowing the storage server to quickly retrieve access information relevant to a particular client. One advantage of using radix tree to maintain access permission information is that radix tree is very efficient at storing hierarchical information, such as IP addresses. Radix tree is also very efficient at representing subnets in particular.

A scheduler utilizes a data structure in the form of an augmented, pruned, radix tree to implement 2-key scheduling.

The invention discloses a dynamic hash table operation method based on a hybrid DRAM-NVM (Dynamic Random Access Memory-Non-Volatile Memory). The method comprises a segment splitting operation, an insertion operation, a query operation, a deletion operation and a recovery operation, the method comprises the following steps: storing an array structure directory for storing a segment address of a hash unit in a dynamic hash table on a DRAM (Dynamic Random Access Memory), and storing a cardinal number tree structure directory for synchronously backing up the array structure directory in the dynamic hash table on an NVM (Non-Volatile Memory) to persist the array structure directory, so that the array structure directory is reconstructed after a system crashes; storing the hash units by adoptinga multi-layer structure, and storing the hash units on the NVM; according to the method, a cross storage key value pair and an atomic segment splitting mechanism are adopted to ensure the consistencyof data, the provided hash table fully utilizes the advantages of a hybrid memory, the expenditure of data consistency is relatively small, and the access delay is very low.

The embodiment of the invention provides a uniform resource locator matching processing method and a uniform resource locator matching processing device. The method comprises the following steps: dividing received uniform resource locators (URL) by levels, and acquiring a second-level domain in each URL, or the second-level domain and information under the second-level domain; hashing the second-level domain to acquire a character string corresponding to the second-level domain, and acquiring a key value according to the character string corresponding to the second-level domain; and when a node corresponding to the second-level domain is matched in a binary search tree (BST) corresponding to the key value, sequentially matching the information under the second-level domain with the corresponding character string in a radix tree under the node corresponding to the second-level domain from high level to low level to acquire a second matching result. According to the method, the second-level domain, a sub-domain, a category, a page and the like under the URL are acquired and sequentially matched, so that domains of all levels under the second-level domain can be matched linearly at one time.

A packet network device includes a route processor that operates to maintain one or more forwarding tables and it includes one or more line cards that operate to process information received by the packet network device from the network and to forward the information to its correct destination. The route processor also operates to identify which incoming prefixes can be used to update the forwarding tables or to identify prefixes stored in the packet network device that can be redistributed from one network protocol to another network protocol running on the route processor. A table management function running on the route processor operates to identify the best match between an incoming prefix and information included in policy statement associated with both an ordered prefix-list and a radix tree structure.

Systems and methods are disclosed for inserting a key into and removing a key from a composite data structure including a radix tree and an auxiliary data structure. In an example, a method of inserting a key into the composite data structure includes receiving an instruction to store a first key in the composite data structure. The composite data structure stores a second key. The method further includes generating, based on comparing the first and second keys, one or more sequences of elements. The method also includes splitting each of the generated sequences of elements into a prefix and a suffix, storing the respective prefixes into the radix tree; and storing the respective suffixes into the auxiliary data structure.

The invention provides a remote dynamic data processing and verifying method and system. The remote dynamic data processing and verifying method comprises providing an n-stage radix tree structure which comprises a root node, a plurality of intermediate nodes and a plurality of leaf nodes; obtaining and recording n initial numerical values to express the empty radix tree structure and designating the identical initial numerical value for the nodes in the same stage; judging the first leaf node in the leaf nodes and computing and recording the numerical values of the nodes on a shortest path from the first leaf node and the root node every time the data processing operation is performed; finding out the second leaf node which is corresponding to specific data in the leaf nodes when the specific data is inspected, obtaining brother nodes of the nodes on a shortest path from the second leaf node to the root node and generating an inspecting result according to number signature, the numerical values of the brother nodes and the specific data which are used for root node verification.

The invention discloses a method used in the I/O of iSCSI target device block equipment. The method comprises the steps that page high speed cache is used and a Radix-Tree-based searching algorithm is adopted between interfaces of the iSCSI target device and the storage equipment. The method aims at enhancing the I/O performance of a storage system, exchanging time by using space, as well as using high-efficient searching algorithm to enhance executing efficiency. The invention adopts the method for improving a software system, basically has nothing to do with the characteristics of hardware equipment and has good universal external interfaces.

The invention discloses a domain name multiline intelligent analyzing method based on a radix tree. According to the method, through adoption of a CLASS field in a DNS protocol family and reservation of an IP set corresponding to a line by use of the radix tree, a brand new BIND9 intelligent analysis scheme is realized. In the novel scheme, each domain name only requires configuration of one zone file and is compatible with old configuration, the IP address of the line can be dynamically increased and decreased, the defect brought about by use of intelligent analysis when bind9 carries big data is well solved, the intelligent analysis is supported under the condition that neither views nor zone files are increased, compatibility with old data is realized, hardware requirements and operation and maintenance technological requirements can be reduced, and the overall system performance is improved.

The invention provides a cardinal number tree access system based on RDMA and a nonvolatile memory. The cardinal number tree access system comprises: a data storage module, wherein the data storage module adopts a nonvolatile persistent memory as a storage medium of a cardinal number tree; a remote read-write module which is used for establishing RDMA communication connection between the server CPU and the client; and a background processing module which is used for carrying out background processing when the client carries out remote reading and writing so as to ensure the consistency of thedata. The method comprises the following steps: S1, in a cardinal number tree initialization stage, constructing a data block to support insertion, update and deletion of far-end data; S2, after the cardinal number tree initial data block is full, reconstructing a data structure into a cardinal number tree structure which comprises a prefix node and a data node, and writing the prefix node of thetree structure into the client; and S3, regularly checking the data nodes of the cardinal number tree. On the premise that data consistency is effectively guaranteed, the pressure of a server-side CPUcan be greatly reduced, the throughput rate is increased, and therefore the overall performance of remote data reading and writing is improved.

The invention discloses a self-adaptive learning type indexing method for workloads in a memory database. According to the method, a cardinal number tree and a piecewise linear model with a maximum error bound are combined, memory occupation of indexes is reduced by utilizing data distribution through a machine learning model, and meanwhile stable query performance is kept. On this basis, an efficient insertion buffer is used to reduce the cost of data insertion update, and in order to alleviate the influence of data insertion on index performance, two workload adaptive recombination optimization methods are used to optimize hotspot data involved in point query and range query in workloads in a targeted manner. The method has relatively high construction efficiency and relatively low memory occupation, relatively efficient query performance is also ensured, and insertion and updating can be well supported; And meanwhile, recombination optimization is performed in a targeted manner by sensing the query workload, so that the influence of insertion on the index performance is reduced at a relatively low cost.

Embodiments of the invention relate to hybrid address translation. An aspect of the invention includes receiving a first address, the first address referencing a location in a first address space. The computer searches a segment lookaside buffer (SLB) for a SLB entry corresponding to the first address; the SLB entry comprising a type field and an address field and determines whether a value of the type field in the SLB entry indicates a hashed page table (HPT) search or a radix tree search. Based on determining that the value of the type field indicates the HPT search, a HPT is searched to determine a second address, the second address comprising a translation of the first address into a second address space; and based on determining that the value of the type field indicates the radix tree search, a radix tree is searched to determine the second address.

Radix trees and other trees use memory inefficiently when storing key-value associations with ‘or’ conditions. Their function can be optimized by using multiple key field trees, each corresponding to a key field, which is typically a character (or group thereof) in a string input key. The tree for the final key field has nodes with the output values, and these are annotated to identify, for each output value, the valid key field values from prior key fields. To execute a lookup, each key field tree is traversed to find a matching key field value. The final key field tree is traversed to reach one or more output values; then the previously determined key field values are compared against the valid key field values to determine if there is a match for a particular output value. The matched and valid key field values can be expressed in encoded form.