Return-Path: Received: from newsfeed.cts.com (newsfeed.cts.com [209.68.248.164]) by naude.co.za (8.11.2/8.11.2) with SMTP id g570Oet31354 for ; Thu, 6 Jun 2002 20:24:40 -0400 Received: from fatcity.UUCP (uucp@localhost) by newsfeed.cts.com (8.9.3/8.9.3) with UUCP id JAA46410; Thu, 6 Jun 2002 09:19:40 -0700 (PDT) Received: by fatcity.com (26-Feb-2001/v1.0g-b71/bab) via UUCP id 004762D0; Thu, 06 Jun 2002 08:03:24 -0800 Message-ID: Date: Thu, 06 Jun 2002 08:03:24 -0800 To: Multiple recipients of list ORACLE-L X-Comment: Oracle RDBMS Community Forum X-Sender: "Michael Rosenblum" Sender: root@fatcity.com Reply-To: ORACLE-L@fatcity.com Errors-To: ML-ERRORS@fatcity.com From: "Michael Rosenblum" Subject: Re: Database link performance Organization: Fat City Network Services, San Diego, California X-ListServer: v1.0g, build 71; ListGuru (c) 1996-2001 Bruce A. Bergman Precedence: bulk Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_0020_01C20D49.5A4EE9D0" ------=_NextPart_000_0020_01C20D49.5A4EE9D0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Hi, all! Thanks for your help and suggestions. Here is some feedback on what I have found on the performance problem of the querying data between two databases using DB links. To John Kanagaraj: filter by ROWNUM does not make any difference - I have absolutely the same speed, when querying table with 5000 rows and filtering 'where ROWNUM <5000'. To Tom Mercadante: creating of a view, that does filtering, did not help. Absolutely the same results! Just remind you the case: declare v_record B%rowtype; cursor cB is select * from B@db_B.mycomp.com where rownum < 5000; begin for c in cB loop v_record :=c; end loop; end; My next step has been to analyze session-level statistics. When using link we have physically two sessions - from a client to db_A (opened explicitly) and from db_A to db_B (opened by Oracle automatically). After execution of a script: - in the session to db_A we have 10000 SQL*Net roundtrips to/from db_link - in the session to db_B we have 10000 SQL*Net roundtrips to/from client. But when I am running the same script directly (from client to db_B) - there is only ONE SQL*Net roundtrip to/from the client. I asked Paul Dorsey to check it out with Tom Kyte at yesterday's NYOUG meeting. Tom told us that Oracle 8.* does single record fetches when using database links. In general, database links have been tuned to support database replication. It means, that 'create table new_A as select . from table_A@db_B' runs quickly but querying from new_A is quite slow. The remaining question: has anybody tried to use database links in Oracle9i? Does the same single fetch limitation apply? Thank you, Michael Rosenblum Dulcian Inc. ------=_NextPart_000_0020_01C20D49.5A4EE9D0 Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Message

Hi, = all!

 

Thanks for your help and suggestions.  = Here is=20 some feedback on what I have found on  the performance = problem=20 of the querying data between two = databases=20 using DB links.

 

To John Kanagaraj: = filter by=20 ROWNUM does not make any difference – I have absolutely the same = speed, when=20 querying table with 5000 rows and filtering ‘where ROWNUM=20 <5000’.

 

To Tom Mercadante: = creating of a=20 view, that does filtering, did not help. Absolutely the same=20 results!

 

Just remind you = the=20 case:

declare =

   v_record = B%rowtype;

 

   cursor cB

   is

   select *

   from=20 B@db_B.mycomp.com

   where rownum <=20 5000;

begin

   for c in = cB

   loop

     v_record=20 :=3Dc;

   end loop;

end;

 

My next step has been = to analyze=20 session-level statistics.

 

When using link we = have=20 physically two sessions – from a client to db_A (opened = explicitly) and from=20 db_A to db_B (opened by Oracle automatically). After execution of a = script:=20
- in the session to db_A we have 10000 SQL*Net roundtrips to/from=20 db_link

- in the session to = db_B we have=20 10000 SQL*Net roundtrips to/from client.

 

But when I am running = the same=20 script directly (from client to db_B) – there is only ONE SQL*Net = roundtrip=20 to/from the client.

 

I asked Paul Dorsey = to check it=20 out with Tom Kyte at yesterday’s NYOUG = meeting.

 

Tom told us that = Oracle 8.* does=20 single record fetches when using database links. In general, database = links have=20 been tuned to support database replication. It means, that ‘create = table new_A=20 as select … from table_A@db_B’ runs quickly but querying = from new_A is quite=20 slow.

 

The remaining = question: has=20 anybody tried to use database links in Oracle9i? Does the same single = fetch=20 limitation apply?

 

Thank = you,

Michael = Rosenblum

Dulcian=20 Inc.

------=_NextPart_000_0020_01C20D49.5A4EE9D0-- -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: Michael Rosenblum INET: mrosenblum@dulcian.com Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051 San Diego, California -- Public Internet access / Mailing Lists -------------------------------------------------------------------- To REMOVE yourself from this mailing list, send an E-Mail message to: ListGuru@fatcity.com (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing).