There is very little consolidated information on how to speed up
import when it is unbearably slow. Obviously import will take
whatever time it needs to complete, but there are some things
that can be done to shorten the time it will take.[@more@]
System Level Changes
- Create and use one large rollback segment, take all other rollback
segments offline. One rollback segment approximately 50% of the size
of the largest table being imported should be large enough. Import
basically does 'insert into tabX values (',,,,,,,')' for every row in
your database, so the rollback generated for insert statements is only
the rowid for each row inserted. Also create the rollback with the
minimum 2 extents of equal size.
- Put the database in NOARCHIVELOG mode until the import is complete.
This will reduce the overhead of creating and managing archive logs.
- As with the rollback segment, create several large redo log files, the
larger the better. Take any small redo logs off line. The larger the
log files, the fewer log switches that are needed. Check the alert
log for messages like 'Thread 1 cannot allocate new log, sequence 17,
Checkpoint not complete'. This indicates the log files need to be bigger
or you need more of them.
- If possible, make sure that rollback, table data, and redo log files
are all on separate disks. This increases throughput by reducing
- Make sure there is no IO contention occurring. If possible, don't run
other jobs which may compete with the import for system resources.
- Make sure there are no statistics on data dictionary tables.
- Check the sqlnet.ora in the TNS_ADMIN location. Make sure that
TRACE_LEVEL_CLIENT = OFF
- Increase DB_BLOCK_SIZE when recreating the database, if possible. The
larger the block size, the smaller the number of I/O cycles needed.
This change is permanent, so consider all effects it will have before
Init.ora Parameter Changes
- Set LOG_CHECKPOINT_INTERVAL to a number that is larger than the size
of the redo log files. This number is in OS blocks (512 bytes on most
Unix systems). This reduces checkpoints to a minimum (only at log
- Increase SORT_AREA_SIZE. Indexes are not being built yet, but any
unique or primary key constraints will be. The increase depends on what
other activity is on the machine and how much free memory is available.
Try 5-10 times the normal setting. If the machine starts swapping and
paging, you have set it too high.
- Try increasing db_block_buffers and shared_pool_size.
Shared pool holds cached dictionary info and things like cursors,
procedures, triggers, etc. Dictionary info. or cursors created on
the import's behalf (there may be many since it's always working on a
new table) may sometimes clog the pipes. Therefore, this stale stuff
sits around until the aging/flush mechanisms kick in on a per-request
basis because a request can't be satisfied from the lookaside lists. The
ALTER SYSTEM FLUSH SHARED_POOL throws out *all* currently unused objects
in one operation, hence, defragments the pool.
If you can restart the instance with a bigger SHARED_POOL_SIZE prior
to importing, that would definitely help. When it starts to slow down,
at least you can see what's going on by doing the following:
SQL> set serveroutput on size 2000;
The dbms_shared_pool is in $ORACLE_HOME/rdbms/admin/dbmspool.sql
Import Options Changes
- Use COMMIT=N. This will cause import to commit after each object (table),
not after each buffer. This is why one large rollback segment is needed.
- Use a large BUFFER size. This value also depends on system activity,
database size, etc. Several megabytes is usually enough, but if you
have the memory some can go higher. Again, check for paging and swapping
at the OS level to see if it is too high. This reduces the number of
times the import program has to go to the export file for data. Each
time it fetches one buffer's worth of data.
- Consider using INDEXES=N during import. The user defined indexes will be
created after the table has been created and populated, but if the primary
objective of the import is to get the data in there as fast as possible,
then importing with INDEXES=N will help. The indexes can then be created
at a later date when time is not a factor.
If this approach is chosen, then you will need to use INDEXFILE option
to extract the DLL for the index creation or to re-run the import with
INDEXES=Y and ROWS=N.
REMEMBER THE RULE OF THUMB: Import should be minimum 2 to 2.5 times
the export time.
Large Imports of LOB Data:
Generally speaking, a good forumla for determining a target elapsed
time for a table import versus the elapsed time for the table export is:
import elapsed time = export elapsed time X 4
- Eliminate indexes. This affects total import time significantly.
The existance of LOB data requires special consideration.
The LOB locator has a primary key that cannot be explicity dropped
or ignored during the import process.
- Make certain that sufficient space in sufficently large contiguous
chunks is available to complete the data load. The following should
provide an accurate image of the space available in the target
alter tablespace mailindx coalesce;
where tablespace_name = 'MAILINDX'
order by bytes desc;
where tablespace_name = 'MAILINDX'
order by bytes desc;
Large Imports of LONG Data:
Importing a table with a LONG column may cause a higher rate of I/O and disk
utilization than importing a table without a LONG column.
There are no parameters available within IMP to help optimize the import of these
来自 “ ITPUB博客 ” ，链接：http://blog.itpub.net/202861/viewspace-799577/，如需转载，请注明出处，否则将追究法律责任。