Online Aggregation: A Review.

Recent demands for querying big data have revealed various shortcomings of traditional database systems. This, in turn, has led to the emergency of a new kind of query mode, approximate query.Online aggregation is a sample-based technology for approximate querying. It becomes quite indispensable in the era of information explosion today. Online aggregation continuously gives an approximate result with some error estimation (usually confidence interval) until all data are processed. This survey mainly aims at elucidating the most critical two steps for online aggregation: sampling mechanism and error estimation methods. As the development of MapReduce, researchers try to implement online aggregation in MapReduce framework. We will also briefly introduce some implementations of online aggregation in MapReduce and evaluate their features, strength, and drawbacks. Finally, we disclose some existing challenges in online aggregation, which needs attention of the research community and application designers.