Two questions [message #182923] |
Tue, 18 July 2006 13:46 |
hshapiro
Messages: 2 Registered: July 2006
|
Junior Member |
|
|
I have 2 questions:
1. What are the advantages to having multiple datafiles in a tablespace, as opposed to having 1 datafile and making sure AUTOEXTEND is on?
2. What is the best way to multiplex the redo logs?
|
|
|
Re: Two questions [message #183533 is a reply to message #182923] |
Fri, 21 July 2006 04:37 |
nmacdannald
Messages: 460 Registered: July 2005 Location: Stockton, California - US...
|
Senior Member |
|
|
1. What are the advantages to having multiple datafiles in a tablespace, as opposed to having 1 datafile and making sure AUTOEXTEND is on?
Using multiple datafiles on seperate disks allows you as the dba to move 'hot' (most active) data to spread the load among disks, allowing the io queue length to be smaller. If you find one tablespace that is overactive then move its supporting disk segments to seperate disks.
2. What is the best way to multiplex the redo logs?
define them in your init.ora file.
archive_log_dest_1 = '<some directory>'
archive_log_dest-2 = '<some other directory on a seperate disk>
|
|
|
Re: Two questions [message #183931 is a reply to message #182923] |
Mon, 24 July 2006 07:49 |
jerrysun
Messages: 3 Registered: June 2006 Location: China
|
Junior Member |
|
|
hi, excuse for my poor english
1、AUTOEXTEND ON will make more costs when there is no enough space left and make the operation on the datafile much more complex when the datafile extend bigger and bigger.
2、multiplex the redo logs on separate disk,speed the LGWR
|
|
|
Re: Two questions [message #183943 is a reply to message #183931] |
Mon, 24 July 2006 08:37 |
JRowbottom
Messages: 5933 Registered: June 2006 Location: Sunny North Yorkshire, ho...
|
Senior Member |
|
|
1) If you've got autoextend on, and the tablespace tries to extend, and you're out of disk space, then you're in EXACTLY the same situation as if you didn't have AUTOEXTEND ON. Anything that requires you to allocate more space to that tablespace is going to error. It isn't going to increase the cost of anything.
It also isn't goint to make operations on the datafile more complex.
If you can explain why you think this is the case, then paerhaps we can explain your misunderstanding.
2) I'm afraid I don't understand your question.
|
|
|