RE: Multiple imports simultaneous?

From: Storey, Robert (DCSO) <"Storey,>
Date: Tue, 29 Dec 2015 21:45:30 +0000
Message-ID: <FE4C2B093843BB4B873D754E5E0BE4DB01998C3DBA_at_DCSOSVMS02.dcso.org>

In this case it would be part of a LUN in a SAN. But, the new hardware I am going to migrate it to (when it gets here) will have 12 disks. I will create 6 mirror pairs, and then stripe across the pairs. 2 separate disks will hold my redo. So, data and indexes will go onto the raid10 where logical volumes will be created just for ease of structure (windows environ).

And my bottleneck has been the table in schema c that has the XMLTYPE as a column. That one table seems to take 12 hours, regardless of how I approach it. I’m still working on speeding it up, or eliminating it from the import.

From: Raza Siddiqui [mailto:raza.siddiqui_at_oracle.com] Sent: Tuesday, December 29, 2015 3:42 PM To: Storey, Robert (DCSO)
Cc: Oracle L
Subject: Re: Multiple imports simultaneous?

Overall # of rows is not huge (in today's terms), or schemas or objects...what is more important is mapping of your tablespaces to their corresponding files to physical disks - that would be your possible bottleneck.

Raza

On Dec 29, 2015, at 13:28, "Storey, Robert (DCSO)" <RStorey_at_DCSO.nashville.org<mailto:RStorey_at_DCSO.nashville.org>> wrote: Just curious about import.

I’m importing an older 9i system into an 11g system. I am precreating the tablespaces, etc.

Ultimately, the import only cares about 3 schemas. One that only has my PL/SQL and the other two have the data. So, Schema A, B, C. Schema B has about 360 tables and about 190 million rows total. Schema D has about 45 tables, but and about 35 million rows of which 27 are in one table that has an XMLTYPE column. Importing just the one table in Schema B takes about 12 hours. I’m working on methods to trim that time.

But, is it possible to do multiple imports at once if using different inbound schemas.

1. Export the database to create my dump file.
2. ON the target server, make 3 copies of the import file.
   1. Do an import of Schema A, rows=n, indexes=n, constraints=n
   2. Do an import of Schema B, rows=n, indexes=n, constraints=n
   3. Do an import of Schema C, rows=n, indexes=n, constraints=n
3. Do an import for each schema where rows =n, and indexes and constraints = y.

Theoretically, this should not interfere with each other. I can set the database to no-archive and increase redo logs so that the waits should be reduced.

Thoughts?

--
http://www.freelists.org/webpage/oracle-l

Received on Tue Dec 29 2015 - 22:45:30 CET