Oracle 10g DBA: Adding Time Information into Tablespace Names

In VLDB databases, there is a common willing in using a tablespace naming convention that includes seasonal information of data within tablespace, such as CHURN_2008M01 or SALE_2008W50. This has various benefits in terms of ILM (Information Life Cycle) management. By just looking the name of a tablespace, the DBA can have a pretty good idea about the tablespace and segments within that tablespace. Nevertheless, prior to 10g this convention can cause significant degeneration in system catalog performance if data windowing is used. In other words, if you periodically need to drop old tablespaces and add new ones, Oracle catalog starts to blow up. In this paper, you will find a tricky way of solving this problem by Oracle 10g.

Read rest of the paper…

This post consists of my article published in IOUG “Oracle Technology Best Practices” booklet delivered in Oracle Open World 2008.

My Open World Presentation Experience

Hi there,

As I’ve promised, I publish my presentation I held on Tuesday in Oracle Open World. It was a really challanging experience  for me to perform a presentation in such a great event. Moreover one of the Oracle experts, Julian Dyke, I really like to read attented to my presentation. And I am very proud of hearing his positive feedbacks about my presentation. Thanks a lot Julian. Hope to see you in Birmingham…

I also want to thank to Reiner Zimmermann from Oracle, all my collegues and managers in Turkcell, and Hasan Tonguc Yilmaz for their supports and ideas in preparing this presentation.

Growing Data Warehouse to 50 TB and beyond…

Oracle 11g SQL: Named & Mixed Notation for PL/SQL in SQL Statements

 In programming language theory, positional notation refers to matching the order of formal parameters and actual parameters.  Whereas named notation refers to matching the formal parameters and actual parameters explicitly usually (not necessarily) by some operator (in PL/SQL this operator is =>). Finally mixed notation refers using both together as a single notation.

Although there was no restriction on using any of those notations within PL/SQL blocks, prior to 11g named & mixed notations for PL/SQL subroutines were not allowed in SQL statements (Code Piece 1 and Code Piece 2). This is one of the most restrictive issues about SQL for developers. Luckily 11g solves that issue and now it is allowed to use named and mixed notations also in SQL statements.

Read rest of the paper…

Oracle 11g PL/SQL: Allow Sequences in PL/SQL Expressions

Yet Another Optimization or a Compiler Trick?

In Oracle 11g, Oracle finally allows PL/SQL developers to use sequences in PL/SQL expressions. Prior to that version developers should fetch the nextval or currval of a sequence by using a select … dual type of statements. By 11g you don’t need this anymore. You can simply refer to sequences anywhere you want within a PL/SQL expression.

In this paper you will find out basic usage example and whether this new syntax is a new PL/SQL optimization removing dual access necessity or it is only plastics.

Read the rest of paper…

“Growing a Data Warehouse to 50 Terabytes and Beyond” in Oracle Open World

I will be in OW this year also. I will be presenting on 22th of September (Monday) in Moscone South / Rm 307. The topic is how I, our  senior VLDB specialist Uğur Demirelçe, and fabulous UNIX/Hardware administrator M. Altuğ Kamaci have rebuild the Turkcell’s Information Factory (IF).

We have recently built up an information organism living on Oracle Technologies for Europe’s 3. GSM operator and world’s 25. communication and technology company Turkcell.

Forget about any documentation, marketing presentations, and newsletters. I will be presenting the real life.In this presentation, I will be sharing our experiences that can help you to keep alive in your own IF environments. I have recently looked up my session and saw that there were less than 250 seats left. So be quick and join me on 22th of September in San Francisco on Oracle Open World…

Oracle 11g RMAN: BZIP2 vs. ZLIB

Prior to 11g Oracle RMAN had a single compression algorithm, called BZIP2. The algorithm has a very satisfactory compression ratio in terms of decreasing the size of RMAN output. However, high CPU cost makes algorithm not suitable for many sites especially for sites having CPU bottleneck (Data warehouse DBAs?!?:)). As a result people still use hardware compression capabilities of tape drivers (ratios like 1:3) to decrease the backup time and increase the effective write speed of backup drivers. By 11g Oracle introduces a new compression algorithm that is announced to be less compressive but less aggressive in terms of CPU. In this paper you will find comparison of two algorithms with no compressed case.

Read the rest of paper…