Re: 20 TB Bigfile
From: Yong Huang <yong321_at_yahoo.com>
Date: Wed, 11 Mar 2015 16:01:13 +0000
Message-ID: <278868063.996443.1426089672997.JavaMail.yahoo_at_mail.yahoo.com>
> there is a completely unimportant matter of backup. A single big > file tablespace means that if anything goes wrong, you have to restore
> 20TB. ... And there
> is a slight difference in speed in restoring 20TB and 32GB.
Very well-said! I want to add that because a bigfile tablespace can only have one datafile (I wish Oracle didn't have this restriction), if your current ASM diskgroup (or filesystem) is full and you have space in another diskgroup, you can't expand the tablespace by adding another datafile. It may just be a small inconvenience if you can quickly add disks to the diskgroup. It's nevertheless a problem especially in emergency. Also, if you decide to bring a datafile offline for recovery (suppose RMAN block recovery is not efficient enough), you have to bring offline the whole tablespace. If you had multiple datafiles and the corruption is quite isolated, you could bring one datafile offline and some SQLs may be lucky to not touch this specific datafile. Basically, a multi-datafile tablespace gives you more options. Yong Huang
Date: Wed, 11 Mar 2015 16:01:13 +0000
Message-ID: <278868063.996443.1426089672997.JavaMail.yahoo_at_mail.yahoo.com>
> there is a completely unimportant matter of backup. A single big > file tablespace means that if anything goes wrong, you have to restore
> 20TB. ... And there
> is a slight difference in speed in restoring 20TB and 32GB.
Very well-said! I want to add that because a bigfile tablespace can only have one datafile (I wish Oracle didn't have this restriction), if your current ASM diskgroup (or filesystem) is full and you have space in another diskgroup, you can't expand the tablespace by adding another datafile. It may just be a small inconvenience if you can quickly add disks to the diskgroup. It's nevertheless a problem especially in emergency. Also, if you decide to bring a datafile offline for recovery (suppose RMAN block recovery is not efficient enough), you have to bring offline the whole tablespace. If you had multiple datafiles and the corruption is quite isolated, you could bring one datafile offline and some SQLs may be lucky to not touch this specific datafile. Basically, a multi-datafile tablespace gives you more options. Yong Huang
-- http://www.freelists.org/webpage/oracle-lReceived on Wed Mar 11 2015 - 17:01:13 CET