Li Fudong (Yi Tsai) Time: July 25, 2014
Since entering the new department, the opportunity to encounter DB Daniel Robin, in the course of the project clicked, Oracle has a lot of insight.
For data processing, the main problem is to solve performance and efficiency problems, because the amount of data is very large, then the use of DDL and DML is shown level. It is important to know when to create, when to insert, and when to update. Big data volume try not to use update, because update relies on indexes, but also needs to ensure rollback and commit, many additional actions, so the general speed is slow. In this case, if you create a temporary table, any insert, and then index the way, is a very good choice. If you use a partition, you can create a new table and a partition, and then swap the new table filled with the data with the new partition, which can improve the speed of data implantation, because the technique of partitioning switching is simply to make a new table-pointing action in the new partition, with the pointer principle being the address change that the pointer is pointing at. Doing so naturally is very fast.
This article is from "Easy to blog" blog, please be sure to keep this source http://lifudong.blog.51cto.com/2357095/1530150