Abstract: This paper studies the problem of Skyline computation within multi-tenant database system and proposes an effective algorithm, multi-tenant database oriented Skyline (MDOS). In multi-tenant database system, data are always stored in pivot table. A pivot table oriented index array based on B+-tree is developed. Based on this index structure, a set of heuristic pruning rules is used, which greatly reduces the times of domination test. Massive experiments demonstrate that MDOS is efficient and is of high overall performance.

Key words: multi-tenant database system, pivot table, Skyline computation, index array, pruning techniques

摘要： 对面向服务的多租户数据库上的Skyline查询展开了深入研究, 提出了一个有效的查询处理算法——面向多租户数据库的Skyline(multi-tenant database oriented Skyline, MDOS)。针对多租户数据库环境下, 数据往往以枢轴表形式存贮的特点, 设计了一个基于B+树的索引阵列。基于该索引结构, 进一步采用启发式剪枝策略, 极大地降低了支配测试的次数。大量的实验表明, MDOS算法具有较高的综合性能。

关键词: 多租户数据库系统, 枢轴表, Skyline计算, 索引阵列, 剪枝技术