 |
(Man Xiong and Sunil Agarwal) Incremental bulk load refers to loading data into a nonempty table. The key question during incremental bulk load is whether indexes should be dropped before bulk load. The answer depends on multiple factors.
This paper attempts to answer this question through a case study of incremental bulk load using the BULK INSERT statement on a representative decision support system (DSS) running Microsoft SQL Server 2000 on Microsoft Windows 2003 Server. You should view the results presented in this paper as recommendations instead of absolute answers, because your results will depend on your hardware (for example, number of CPUs, I/O, and network bandwidth), data distribution in the table, and the data files and parameters supported by the BULK INSERT statement. All tests were conducted on a Hewlett-Packard hardware system (for more information about the test environments, see Appendix A, "Test Environments"). |
