China
Africa
Explore chapters and articles related to this topic
Databases
Published in Ian Foster, Rayid Ghani, Ron S. Jarmin, Frauke Kreuter, Julia Lane, Big Data and Social Science, 2020
A second way in which the user can contribute to performance improvement is by using appropriate table definitions and data types. Most DBMSs store data on disk. Data must be read from disk into memory before they can be manipulated. Memory accesses are fast, but loading data into memory is expensive: accesses to main memory can be a million times faster than accesses to disk. Therefore, to ensure queries are efficient, it is important to minimize the number of disk accesses. A relational DBMS automatically optimizes queries: based on how the data are stored, it transforms a SQL query into a query plan that can be executed efficiently, and chooses an execution strategy that minimizes disk accesses. Even so, users can contribute to making queries efficient. As discussed, the choice of types made when defining schemas can make a big difference. As a general rule, only use as much space as needed for your data: the smaller your records, the more records can be transferred to main memory using a single disk access. The design of relational tables is also important. If you put all columns in a single table (i.e., you do not normalize), more data may come into memory than is required.
Multi connection query optimization in data warehouse dependent on multiple linear regression algorithm
Published in International Journal of Computers and Applications, 2019
If several queries cannot be shared, it is clear that the overall cost of the MCQODW strategy is the smallest. And it can be seen that for the registration and deletion of queries, Algorithms 1 and 2 can easily adjust the query plan. From the discussion in Figure 1, it can be known that there are two kinds of methods for the multi connection query optimization of the data warehouse: (1) n-ary DR, that is, whether to update the query plan is determined in accordance with the adjusted cost; (2) MCQODW strategy. Obviously, when the weight is close to the first method, relatively good performance can be obtained. In the next section, the worst case performance under the MCQODW strategy is analyzed in two scenarios as the following.