Row chaining representing in redo log
From: SY Wang <like.shaq_at_gmail.com>
Date: Mon, 30 Nov 2009 09:43:38 +0800
Message-ID: <f705dc2d0911291743n36281b70xbcf208a56bff79d6_at_mail.gmail.com>
Currently, I am working on redo logs, and I found it is possible to understand the row chaining by dumping the redo log file, but it is hard to make the algorithm or logical strategy that Oracle does, which I mean how Oracle implement row chaining.
Date: Mon, 30 Nov 2009 09:43:38 +0800
Message-ID: <f705dc2d0911291743n36281b70xbcf208a56bff79d6_at_mail.gmail.com>
Currently, I am working on redo logs, and I found it is possible to understand the row chaining by dumping the redo log file, but it is hard to make the algorithm or logical strategy that Oracle does, which I mean how Oracle implement row chaining.
For example, Suppose a transaction contains two DML operations, both of them need row chaining operation, and How could I separate the two records? I mean they are marked with same transaction id(xid, of course), same bdba and hdba value. How do we know that the row chaining of the first record are finished?
--
Black Thought
--
http://www.freelists.org/webpage/oracle-l
Received on Sun Nov 29 2009 - 19:43:38 CST