CSC5120 Course Project Homepage

Survery on view materialization and indexing method for OLAP and warehouse maintenance problem

[home][intro][progress][document][links][people][FAQ]
FAQ

What is data warehousing?

Data warehousing refer to the collection of business related data of an enterprise throughout a long period, and this data is use to support decision making by mangerial staff of the corporations. top

What is OLAP?

OLAP stands for OnLine Analytic Processing. This is a time-critical process, and the query usually for trend analysis which may consist of numerous aggregation and joining across many table, which means time-consuming, so the difficulties is to speed up such query. top

Why OLAP/data warehousing considered different from traditional database management system/OLTP (OnLine Transaction Processing)?

Since data warehousing is for storage of historical information of the enterprise, so the space requriement is different from traditional DBMS, and may be up to several TB. For the OLAP, it is mainly query with many aggregation and join table operatio (e.g. query for the revenue of each outlet , however, the OLTP, consists of mainly update of a small set of data (e.g. update one row in a table for an business transaction). As a result, we can see that we have to optimize these two application differently, so result in a lot of research. top
What's mean by materialization?

Materialization means make some thing comes into existence. In database terminology, materialization means storing some precomputed queries, in order to speed up other/similar queries by resuing the precomputed result. (e.g. we keep a table containing monthly sales figure by each outlet, so we can answer queries like "the yearly sales figure of each outlet, or the total monthly sales figure by all outlets). top

What is the difficulties in choosing views to materialize?
Materializing views is to speed up the on-line query (described in previous question). However, materializing all views takes up space, and the update of view also takes times. So, this is the question of how to optimize the query time with a space and update time constraint. Moreover, some views can be computed from other views quickly, we also need to avoid such views as materializing it give little benefits to us. top

[home][intro][progress][document][links][people][FAQ]
Copyright 2001