Side-Effect Estimation: A Filtering Approach to the View Update Problem

Abstract

Views and their updates have long been a fundamental technology required in a wide range of applications. However, it has been known that updates through views is a classical intractable problem. In this paper, we propose a novel, data-oriented approach to this problem that provides a practical support for view updates. In particular, we propose a summarization of the source database of views, which serves as an update filter. The update filter aims to efficiently reject untranslatable view updates by estimating the side effects of the updates, thereby avoiding costly translation analysis. For applications where estimation errors are not preferred, our update filter can be tuned to be exact. In this paper, we present our approach with SPJ views, an important class of view definitions. We first revise the notion of estimation errors to quantify the filter's qualities. We then propose a novel join cardinality summary (JCard) derived from cardinality equivalence. An estimation algorithm is proposed. Finally, we present optimizations enabling the construction of an accurate JCard through heuristics and sampling. Our extensive experiments show that update filters are efficient and can be easily tuned to produce accurate estimations on TPC-H and DBLP.

Lossless Selection Views under Conditional Domain Constraints

“…T HE problem of updating a database through a set of views consists in propagating updates issued on the views to the underlying base relations over which the view relations are defined, so that the changes to the database reflect exactly those to the views. This is a classical problem in database research, known as the view update problem ( [1], [2], [3]), which in recent years has received renewed and increasing attention ( [4], [5], [6], [7], [8], [9]).…”