1, exp direct=y recordlength=65535
IN direct mode, exp does comparable performance as expdp. Just the dump file larger round 15% .
2, imp commit=y recordlength=65535 buffer=10485760
imp does the essentially different way as impdp .
For imp, it always does a convertional import, Never direct load.
For impdp, it does the direct load.
So the test result is as expected: impdp is much faster than imp. 2 times faster or Evern great.
-- parameter of imp which about index
indexes=yes => imp execute index ddl , if the index is alread exists. Then skip it.
indexes=no =>imp does not execute index ddl .
skip_unusable_index=yes
If the index is alread exists and with unusable status, and if this parameter is not specified by imp . Then imp raise an error, the table data about the index will not be loaded. The default action about unusable index of imp is different with impdp . which(impdp) will load data , and stay index in orginal state or replace by DDL in dumpfile.
[@more@]来自 “ ITPUB博客 ” ,链接:http://blog.itpub.net/350519/viewspace-1042213/,如需转载,请注明出处,否则将追究法律责任。
转载于:http://blog.itpub.net/350519/viewspace-1042213/
本文通过实验对比了Oracle数据库中EXPDP与IMPDP两种数据导入导出工具的性能。结果显示,在直接加载模式下,EXPDP与IMPDP的性能相近,但IMPDP产生的文件大小更大约15%。此外,文章还详细分析了IMP与IMPDP的工作方式及参数设置,发现IMPDP在数据加载速度上比IMP快两倍。
20

被折叠的 条评论
为什么被折叠?



