内存数据库系统SwiftMMDB索引机制的设计与实现

摘要

T树结合了平衡二叉树(AVL树)和B树的优点,可以有效地组织索引数据,从而为内存数据库提供优良的存储效率和查询性能.结合自主开发的一个内存数据库系统SwiftMMDB介绍T树索引的设计与实现,并通过节点分裂、填充等方法改进了经典T树的插入和删除操作,减少了T树中平衡旋转的次数,从而进一步提高内存数据库检索的效率和性能.%T tree, combining the advantages of AVL tree and B tree, can organize index data efficiently, thus providing good storage efficiency and search performance for main memory database. The design and implementation of T tree index was presented in the main memory database system named SwiftMMDB which was developed by the authors' research group recently. The insert and delete operations of traditional T tree were improved through node splitting and node populating method. The rotation number for balance in T tree was reduced. As a result, the retrieve efficiency and performance of main memory database system are improved.