Framework For Data-intensive Applications Research Articles

A novel virtual disk bandwidth allocation framework for data-intensive applications in cloud environments

A novel virtual disk bandwidth allocation framework for data-intensive applications in cloud environments

A novel virtual disk bandwidth allocation framework for data-intensive applications in cloud environments

Recently, cloud computing has become a promising distributed processing paradigm to deploy various kinds of non-trivial applications. In those applications, most of them are considered data-intensive and therefore require the cloud system to provide massive storage space as well as desirable I/O performance. As a result, virtual disk technique has been widely applied in many real-world platforms to meet the requirements of these applications. Therefore, how to efficiently allocate the virtual disk bandwidth has become an important issue that needs to be addressed. In this paper, we present a novel virtual disk bandwidth allocation framework, in which a set of virtual bandwidth brokers are introduced to make allocation decisions by playing two game models. Theoretical analysis and solution are presented to prove the effectiveness of the proposed game models. Extensive experiments are conducted on a real-world cloud platform, and the results indicate that the proposed framework can significantly improve the utilisation of virtual disk bandwidth comparing with other existing approaches.

High-performance XML modeling of parallel queries based on MapReduce framework

With the increasing of data at an incredible rate, the development of cloud computing technologies is of critical importance to the advances of researches. MapReduce is a widely adopted computing framework for data-intensive applications running on clusters. Traditional parallel XML parsing and indexing approaches are inadequate for processing large-scale XML datasets on clusters and; therefore, we propose an approach to exploit data parallelisms in XML processing using MapReduce in Hadoop. Our solution seamlessly integrates data storage, labeling, indexing, and parallel queries to process a massive amount of XML data. Specifically, we introduce an SDN labeling algorithm and a distributed hierarchical index using DHTs. More importantly, we design an advanced two phase MapReduce solution that is able to efficiently address the issues of labeling, indexing, and query processing on big XML data. The first MapReduce phase applies filtering, labeling, index building techniques, in which each DataNode performs elements labeling using a map function and a reduce function to merge and build indexes. In the second phase, local XML queries in multiple partitions are performed in parallel using index-table-enabled B-SLCA. Our experimental results show the efficiency and effectiveness of our proposed parallel XML data approach using MapReduce Framework.

Efficient Querying Distributed Big-XML Data using MapReduce

MapReduce is a widely adopted computing framework for data-intensive applications running on clusters. This paper proposed an approach to exploit data parallelisms in XML processing using MapReduce in Hadoop. The authors' solution seamlessly integrates data storage, labeling, indexing, and parallel queries to process a massive amount of XML data. Specifically, the authors introduce an SDN labeling algorithm and a distributed hierarchical index using DHTs. More importantly, an advanced two-phase MapReduce solution are designed that is able to efficiently address the issues of labeling, indexing, and query processing on big XML data. The experimental results show the efficiency and effectiveness of the proposed parallel XML data approach using Hadoop.

Query-Driven Frequent Co-Occurring Term Computation over Relational Data Using MapReduce

Given a keyword query q and a large structured, traditional keyword search may return a large number of relevant results to users, which leads to a frustrating procedure for the users to select their interesting results. To help users understand the data to be searched, in this work we investigate the problem of frequent co-occurring terms (FCTs) in large relational data. By returning a set of most FCTs with the given keywords, we can provide a chance for users to see a big picture of relevant data information. The investigation of FCT problem is also one of the fundamental building blocks of data mining because the discovered FCTs can be employed to analyze the topics or contexts of user interest. Although the problem of FCTs computation was proposed and investigated in Tao and Yu [(2009) Finding Frequent Co-Occurring Terms in Relational Keyword Search. 12th Int. Conf. Extending Database Technology EDBT, Saint-Petersburg, Russia, March 23–26, pp. 839–850. ACM, New York, USA], further investigation is needed to improve the performance because FCT computation is very expensive. Especially for the increasing volume of data, the centralized approach in Tao and Yu [(2009) Finding Frequent Co-Occurring Terms in Relational Keyword Search. 12th Int. Conf. Extending Database Technology EDBT, Saint-Petersburg, Russia, March 23–26, pp. 839–850. ACM, New York, USA] may incur a big challenge on the efficiency of performing an FCT computation. To do this, we investigate how to perform parallel FCT computation using MapReduce which is a well-accepted framework for data-intensive applications over clusters of computers. We design an effective mapping mechanism that exploits the approximately maximal workload of FCT computation for balancing the computational cost of each processor, while reducing the shuffling cost and avoiding the data-skewness. Analytical and experimental evaluations demonstrate the efficiency and scalability of our proposed approach using TPC-H benchmark datasets with different sizes.