|
Database File Naming
There have been many different naming conventions
used over the years by Oracle and others for naming the various Oracle
database data files. There are only three basic database file types:
control files, redo log files, and database files. The OFA specified
naming guidelines, and specifically OFA rule 6 specified the naming and
format. Unfortunately the guidelines for redo log naming were poorly
chosen by terminating the file with a ".log" extension. This is easily
confused with the many text log files created by various utilities. This
is the only area where we diverge from OFA "examples". However we still
meet rule 6's requirements along with the spirit of the OFA's naming
guidelines. Therefore the following general naming conventions should be
followed and will be detailed in the following sections:
|
File Type |
Format |
Notes |
|
control file |
control[mm].ctl |
mm – 01,02,… |
|
redo logs |
log[mm][n].rdo |
mm – 01,02,… (group);
n – a,b,c,… (member) |
|
data files |
[tablespace name][TAB|IDX|LOB][mm].dbf |
mm – 01,02,… |
|
archive redo logs |
[SID]_%t_%s.arc
10g: [SID]_%t_%s_%r.arc |
SID – Database SID, %t – thread, %s – log
sequence, %r – reset logs ID (10g only) |
Control files are the way the Oracle database
keeps track of structural information about database files, database
consistency and backup information and are therefore critical to the
basic functioning of the database. It is strongly recommended that at
least three copies of the control files be maintained (e.g. OFA
guidelines as well) and that they be preferably kept on different
physical volumes. The format for naming control files is:
control[mm].ctl where mm is a two digit
sequentially incrementing number starting with 01. Three control files
would then be named:
·
control01.ctl
·
control02.ctl
·
control03.ctl
Oracle redo log files record information about
all database changes and are critical to database recoverability from
both instance failure and media failure. As we have said, the OFA naming
guidelines here fail to take into account that files with a ".log"
extension are generally considered to be informational text based files
and not critical database recovery files. We feel the following is a
better naming guideline that helps identify the type of file and
accommodates multiple groups and group members:
log[mm][n].rdo where mm is a two digit
sequentially increasing number starting with 01 and representing the
redo log file group number, and n is a sequentially increasing letter
starting with a and representing the redo log file group's member.
The following example shows three redo log files
groups with two members each:
|
Group 1 |
Member 1 |
log01a.rdo |
|
|
Member 2 |
log01b.rdo |
|
Group 2 |
Member 1 |
log02a.rdo |
|
|
Member 2 |
log02b.rdo |
|
Group 3 |
Member 1 |
log03a.rdo |
|
|
Member 2 |
log03b.rdo |
Closely related to redo log file format is
archivelog file format. Archivelog files are created by the ARCH process
from redo log files once they have filled if the database is in
archivelog mode. The naming format is controlled by the init.ora
parameter "log_archive_format" and the following is our recommended
format:
LOG_ARCHIVE_FORMAT = [SID]_%t_%s.arc
where [SID] is the database system identifier.
In Oracle 10g the format changes
slightly with the addition of the %r parameter:
LOG_ARCHIVE_FORMAT = [SID]_%t_%s_%r.arc
where [SID] is the database system identifier.
|
