Import takes long time

[iris]

hi, why does import takes longer time compared to export?


thanks,

[davey23uk]

have you thought about what it is actually doing compared to export

[iris]

Hi, In my opinion import takes long time due to following factors:

1. enable constraints
2. creates tables
3. creates indexes
4. executes procedures, functions.

Iam I right ? Please confirm.


thanks,

[PAVB]

It does not executes procedures, functions.

[syedfarukhali]

hi, why does import takes longer time compared to export?


thanks,

Hi,


It takes longer time because of indexes.
So Import the database with parameter indexes=no
import will be fast after that import the indexes.

[PAVB]

...after that import the indexes.

huh? Did you mean "build" the indexes after?

Also, be sure you do not import constraints. Import would try to validate them so is going to be faster if you create constraints after import is complete including the NOVALIDATE parameter.

[abhaysk]

Hi,


It takes longer time because of indexes.
So Import the database with parameter indexes=no
import will be fast after that import the indexes.

Import will anyway build indexes after dataload.. then what's the point?.. only thing that doesnt do is "parallel".. for that use expdp/impdp..

Abhay.

[syedfarukhali]

Import will anyway build indexes after dataload.. then what's the point?.. only thing that doesnt do is "parallel".. for that use expdp/impdp..

Abhay.

While importing with param index=no indexes will not load in the database
after data load in the database we can give the param indexfile to create the indexes

2)If you are really in Hurry give commit=no while importing it will load the data fast but make sure your rollbaack segment must be very huge.

[Mr.Hanky]

While importing with param index=no indexes will not load in the database
after data load in the database we can give the param indexfile to create the indexes

2)If you are really in Hurry give commit=no while importing it will load the data fast but make sure your rollbaack segment must be very huge.

Be carefull with using no commit, if the table is very large you will blow out your undo and then that table import will rollback and you will have to re-import it, and that sucks.

I prefer using a buffer and commit to avoid that.

COMMIT=Y
BUFFER=12000000

Something like that helps, you will also avoid the rare instance (IME) where the default buffer is not large enough to swallow the longest column in the table, I've seen that once or twice.

[gopu_g]

you can go for datapump...if you are using the oracle10g.

Other wise you can follow this steps as the above person said

COMMIT=Y
BUFFER=12000000
just like commit=y you can use nolog=y
direct=y

or

you can use incremental export

at the time of exporting you can export the indexes in to another file...
you can get lot of solution in the Google.... :-)