A Unix Perspective on Oracle Archive Redo Log Files
Mark Bole
I am keenly interested in how organizations define and manage the respective
roles of the Unix Systems Administrator (SA) and Oracle Database Administrator
(DBA). Many reasonable people will maintain that it is unlikely to find one
person who can perform both jobs well, while others have enjoyed great success
in environments where there is shared responsibility for both the root password
and the SYSTEM password. Auditing concerns, career paths, on-call schedules,
and egos further cloud the issue, sometimes leading to the construction of brick
walls where perhaps a picket fence or even a chalk line on the ground would
be more appropriate.
Those who are not familiar with the latest version of Oracle's flagship product,
Database 10g, may be surprised to find that the distinction between SA and DBA
is becoming even less clear. The very first document you see is called "2 Day
DBA", which is targeted to someone with no previous DBA experience, such as
an SA, and which promises to cover all the basic tasks one typically performs
in support of Oracle 10g for a small- or medium-sized organization. Also, Oracle
now provides many features that previously may have been considered the sole
province of the SA, such as:
- Logical disk volume manager (for all database files)
- Cluster file system and clusterware management that requires no third-party
software
- Lempel-Ziv compression (like gzip)
- Copy any operating system file locally or between servers
- Job scheduling capabilities, including jobs completely external to the
database, that vastly exceed those of cron
- POSIX-compliant regular expression handling
Nevertheless, one specific feature of Oracle has been in place from the beginning
-- the archived redo log file, which is arguably the single most important component
of a production application for the SA and DBA to jointly manage and protect.
|
