| With the rapid development of computer and database technology, humanitysociety has entered the age of information; the data that needs to be stored hasincreased considerably, far more than the tolerance range of a single server’scapability. For the requirement of retrieving the data, large index system often basedon a distributed system. But in some cases which demand low latency and highflexibility for handling the response to the scene, the distributed system has theessential of difficulty. Consequently, the improving of storage and processingcapacity has irreplaceable significance especially for the high-profile servers.For the feature of modern server hardware architecture and memory resources,this paper proposes a memory indexing data structures LC-Tree. For CPU cache,branch prediction, memory false sharing and other hardware features to adjust andoptimize LC-Tree data structure and memory layout. Through constructing256-branch tree in logic as the upper structure, the branch node structure use bitmapindexes, immediately indexes and other way to quickly locate the bottom node.Sequential location of bottom leaf node in memory layout has advantage for datacompression algorithm to save limited memory resources.For memory resource scarcity, index compression and encoding can effectivelysave storage space and reduce I/O traffic and increase system throughput. Thedisadvantage of compression is that the compression of the index requiresdecompression during the index operations, such as query, update, delete, etc., willtake additional time and space complexity. This paper presents a memory indexingdata structures LC-Tree and compression storage solutions effectively incompression ratio, decompression achieve trade-off between time and dynamicperformance by implementing dynamic update to support real-time data updates. Forthe requirement of large-scale distribution system this paper presents a scalablesolution in order to satisfy the demands of large data retrieval of data. |
PDF Full Text Request