An Enterprise Dataspaces Visual Fusion Flow Modeling Method And Query Optimization Strategy

During the process of information construction, enterprises build business systems gradually, and accumulate large amounts of business data. How to use these data to provide enterprises with high value-added information is a hot topic of research. However, during the construction of business systems, building concept, development of technology, business characteristics, human, and many other factors, business systems using different data models, different representations of business data representation, leading to heterogeneous data problems between business systems. From enterprise business needs perspective, some applications need to access data in multiple business systems, and integrate data from multiple datasources, including some real-time requirements, such as completing the form. Traditional data integration strategies encounter difficulties in resolving these problems, for example, datawares can't provide real-time queries. We introduce enterprise dataspaces to manage enterprise datasources, providing on-demand, real-time, and flexible queries.This research project mainly to shorten the time that user gets data from enterprise dataspaces, and has two aspects, first, we help user build queries easily, secondly, we shorten the respose time of queries.The main contributions of the thesis are as follows:●Visual fusion flow modeling tookit.It will take a long time to learn how to construct a real-time and correct query. So we develop a visual modeling tool to build queries by way of dragging and dropping. User can save more than 80% time which they spend on learning how to construct queries and building queries on demand through simple dragging and dropping. There is also one example to show convenience of visual modeling, and have been used in the project.●Caching technology application in Enterprise Dataspaces.Enterprise dataspaces contains a variety of heterogeneous and distributed datasources. It has strong network's reliance when we make queries, some queries need long time calculation, and there is some delay. These factors will result in long respose time, so we need to improve its query performance. Caching is an effective way to improve performance. For some frequently used queries, we can reduce the response time by means of caching query results with Key-Value style, and refresh the cache periodly through updating query result.●These methods have been designed and realized.Experiments are designed based on supporting environment "XBUS data aggregation platform". Compared with not using cahce, the experiments results showed that these strategies which combine user confinging and cache refreshing periodly are applied to Enterprise Dataspaces. Experimental results show the average query response time will reduce 30% or more, basically meet the requirements of query response time. At the same time, caching will occupy some memory space, and cache refreshing will make CPU use ratio increase.