Does import use the buffer cache?

[johnharry]

Dear all,

I am trying to get a definate answer to this question; Does the import utility use the database buffer cache when being used?
I know there is an import parameter buffer but I am wondering if you can improve the performance of an import by tuning the buffer cache or is this irrelevant?

Thanks in advance,

John

[akhadar]

u can improve the import performance by increasing the BUFFER value but too much will leads to OS paging and swapping.

[johnharry]

My question is, 'does import use the buffer cache?'
Any takers?

[julian]

The export parameter BUFFER specifies the size in bytes of the buffer used to fetch rows from a table. This has nothing to do with the buffer cache.

[tamilselvan]

The answer is YES.
Import uses buffer cache in the SGA.

SQL> @sga_usage

Before IMPORT

Block Size WHAT        Size in MB
---------- ----------- ----------
      8192 Used                12
      8192 Free Memory       1068
     32768 Free Memory         94


AFTER IMPORT

Block Size WHAT        Size in MB
---------- ----------- ----------
      8192 Used               288
      8192 Free Memory        792
     32768 Free Memory         94 

Tamil

[Mr.Hanky]

Originally posted by julian
The export parameter BUFFER specifies the size in bytes of the buffer used to fetch rows from a table. This has nothing to do with the buffer cache.

Then where is the "BUFFER" space allocated?

[hacketta1]

From the 8i docs:
======
Conventional path Export uses the SQL SELECT statement to extract data from tables. Data is read from disk into a buffer cache, and rows are transferred to the evaluating buffer. The data, after passing expression evaluation, is transferred to the Export client, which then writes the data into the export file.


Direct path Export extracts data much faster than a conventional path export. Direct path Export achieves this performance gain by reading data directly, bypassing the SQL command processing layer and saves on data copies whenever possible.


Figure 1-2 shows how data extraction differs between conventional path Export and direct path Export.


In a direct path Export, data is read from disk into the buffer cache and rows are transferred directly to the Export client. The evaluating buffer is bypassed. The data is already in the format that Export expects, thus avoiding unnecessary data conversion. The data is transferred to the Export client, which then writes the data into the export file
=======
So the buffer cache is always used.

My Hanky, It is my understanding that the evaluating buffer is grabbed from the OS, hence why you sometimes read warnings about not setting the buffer parameter so high that you start to page. Happy to be corrected here.

[johnharry]

This thread is about IMPORT not EXPORT!!!!!

[akhadar]

Did u read tamil's response?

[tamilselvan]

I think some of you have confusion about "buffer cache in SGA" and buffer parameter used in imp utility.

The buffer parameter used in imp is the size for the data allocated in OS Memory for the imp utility. Imp uses this memory area for storing the data before it sends to oracle.

During the imp process, oracle receives data from the imp buffer and reconstructs the data according to its own format before writing into the actual blocks. The writing of dirty blocks is done DBWn process.

Tamil