Redo I/O size seems much bigger the data write I/O
From: amihay gonen <agonenil_at_gmail.com>
Date: Wed, 6 Aug 2014 15:28:46 +0300
Message-ID: <CAKb+SBXBv9UMh=FeLL-+9iZeKQCVBFbyLiEa6k9Ab1t9sBH56Q_at_mail.gmail.com>
Hi all,
in our test machine I see that the ratio between Redo size 1M per second is much lower then the Physical write 78.1 * 8K = 624K
Date: Wed, 6 Aug 2014 15:28:46 +0300
Message-ID: <CAKb+SBXBv9UMh=FeLL-+9iZeKQCVBFbyLiEa6k9Ab1t9sBH56Q_at_mail.gmail.com>
Hi all,
in our test machine I see that the ratio between Redo size 1M per second is much lower then the Physical write 78.1 * 8K = 624K
I wonder if it is because our load test program , or also in a typical oracle production , we'll see more redo the Physical writes .
Assuming the following :
1. no direct insert (bypassing oracle)
2. DB is in archive log (production ...)
Load Profile
Per Second Per TransactionPer Exec Per CallDB Time(s): 0.80.0 0.0010.03 DB CPU(s):0.7 0.00.00 7.97Redo size: 1,685,501.92,699.9 Logical reads: 8,403.5 13.5 Block changes: 13,908.422.3 Physical reads:0.1 0.0 Physical writes: 78.10.1 User calls:0.1 0.0 Parses: 0.60.0 Hard parses:0.1 0.0 W/A MB processed: 0.00.0 Logons:0.0 0.0 Executes: 12,495.320.0 Rollbacks:0.0 0.0 Transactions: 624.3
-- http://www.freelists.org/webpage/oracle-lReceived on Wed Aug 06 2014 - 14:28:46 CEST