基于商立方体的OLAP缓存研究与实现

发布时间：2018-08-13 19:12

【摘要】：在当今信息爆炸时代,电信、银行、在线电子商务网站等数据密集型企业积累了大量的数据,为了分析和利用这些数据,有效利用客户端的缓存,使用户查询效率更高。当今存在许多的查询缓存技术：如语义缓存,对象缓存等查询缓存,语义缓存是指将查询结果和相关查询语句同一存储在缓存中。本文提出一种新的缓存技术-基于商立方体的OLAP缓存模型。该缓存模型优点是：不仅在缓存中存储查询结果,而且可以避免冗余存储,节省缓存空间。同时与其它的缓存技术相比,使用该缓存模型进行查询,需要更少的响应时间,有更高的命中率。本文首先提出一种基于商立方体的OLAP缓存模型,然后对该模型进行系统实现,最后用该模型进行详细的实验。实验结果充分证明了该缓存模型的有效性和准确性。本文的主要工作有： (1)根据用户提交的不同查询,在缓存中动态地生成商立方体的等价类。用户每一次提交的查询首先会在缓存中进行查询,当缓存中不存在与查询项是包含匹配关系等价类的时候,会到数据库中进行查询,然后根据聚集函数得到相对应的聚集值,最后在缓存中生成商立方体的等价类； (2)对整体性度量函数和非单调度量函数进行商立方体的等价类生成。这是本文的第一个创新点； (3)研究利用等价类之间的上卷和下钻的关系,通过剪枝等方法进一步提高缓存查询效率。由于等价类的上界之间存在上卷和下钻的关系,可以将这些等价类组合成一个代数格。这是本文的第二个创新点； (4)研究缓存命中率和响应时间、缓存大小、缓存替换策略以及用户提交查询之间的关系,对OLAP缓存进行测试和验证。
[Abstract]:In the era of information explosion, many data-intensive enterprises, such as telecom, bank, online e-commerce website, have accumulated a lot of data. In order to analyze and utilize these data, the cache of client can be used effectively, and the query efficiency of users will be higher. Nowadays, there are many query caching techniques, such as semantic cache, object cache and so on. Semantic cache is to store query results and related query statements in the same cache. In this paper, a new caching technique, OLAP caching model based on quotient cube, is proposed. The advantage of this caching model is that it not only stores query results in cache, but also avoids redundant storage and saves cache space. Compared with other caching techniques, the cache model requires less response time and higher hit ratio. In this paper, a OLAP cache model based on quotient cube is proposed, and then the model is implemented systematically. Finally, the model is used to carry out detailed experiments. The experimental results demonstrate the validity and accuracy of the cache model. The main work of this paper is as follows: (1) according to the different queries submitted by the user, the equivalent classes of the quotient cube are dynamically generated in the cache. The query submitted by the user is first queried in the cache. When there is no matching relation class in the cache, the query is made to the database, and the corresponding aggregation value is obtained according to the aggregation function. Finally, the equivalent classes of quotient cubes are generated in the cache. (2) the equivalence classes of quotient cubes are generated for integral metric functions and non-monotone metric functions. This is the first innovation point of this paper. (3) the relationship between the equivalent classes is studied to further improve the efficiency of cache query through pruning and other methods. Because of the relation between the upper bound of the equivalence class and the bottom drill, these equivalent classes can be combined into an algebraic lattice. This is the second innovation of this paper. (4) the relationship between cache hit rate and response time cache size cache replacement strategy and user submitted query is studied to test and verify the OLAP cache.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP333

【参考文献】