01.09.2016 Views

Beginning Oracle Database 11g Administration From Novice to Professional

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

344<br />

CHAPTER 15 THE BIG PICTURE AND THE TEN DELIVERABLES<br />

at least one year if you can afford the space. Baseline snapshots should be retained<br />

indefinitely. For example, you can designate the period between 9 a.m. and 10 a.m.<br />

every Monday morning as a baseline period so that the snapshots marking the<br />

beginning and end of the period are retained indefinitely.<br />

9. Exception reports: This deliverable includes reports on SLA violations, security violations,<br />

backup failures, and the like. For example, a certain s<strong>to</strong>red procedure or<br />

SQL statement may have been identified as critical <strong>to</strong> the business, and an exception<br />

report can be produced by mining STATSPACK data.<br />

10. Audit reports: This deliverable typically refers <strong>to</strong> audit reports conducted by<br />

security audi<strong>to</strong>rs but can also refer <strong>to</strong> internal audits of compliance with organizational<br />

processes such as Change Management or database reviews by external<br />

consultants. The absence of audits indicates a lack of oversight of the database<br />

management function.<br />

The Book You Really Need and the Art of the SOP<br />

The book you really need will never be found in books<strong>to</strong>res—it is the book containing all<br />

the procedures that you need <strong>to</strong> operate your databases. You’re going <strong>to</strong> have <strong>to</strong> write<br />

that book yourself. Nobody can write the book for you, because you have a unique environment<br />

and nobody except you would write a book that caters <strong>to</strong> a unique environment.<br />

Do you know how <strong>to</strong> start or s<strong>to</strong>p a database? I thought I did—until I went <strong>to</strong> work<br />

in a large Network Operations Center. We had Solaris, AIX, HP/UX, Linux, and Windows.<br />

We had <strong>Oracle</strong> 8i, <strong>Oracle</strong> 9i, and <strong>Oracle</strong> 10g. We had VCS, HP Service Guard, Sun Clusters,<br />

Data Guard, and RAC. There were so many variations of the startup and shutdown procedures<br />

that I could not remember all of them.<br />

A common task such as adding a data file <strong>to</strong> a database requires different methods<br />

depending on whether you are using cooked files, raw devices, or ASM. Additional complexities<br />

are introduced by RAC and Data Guard. Raw files in particular are no<strong>to</strong>riously<br />

difficult <strong>to</strong> manage—they make it easy <strong>to</strong> damage the database. And, in my experience,<br />

database administra<strong>to</strong>rs routinely forget the important step of backing up the data file<br />

immediately after it is created.<br />

A TRUE STORY<br />

A database administra<strong>to</strong>r s<strong>to</strong>pped a database using the command in preparation for moving<br />

some data files <strong>to</strong> a new location. Unknown <strong>to</strong> him, the database was managed by Veritas high availability<br />

software, which au<strong>to</strong>matically restarted the database. The database was corrupted when the DBA<br />

moved the files.<br />

It happened <strong>to</strong> me—it could happen <strong>to</strong> you.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!