Comments on: The Core Performance Fundamentals Of Oracle Data Warehousing – Introduction

By: The Core Performance Fundamentals Of Oracle Data Warehousing – Balanced Hardware Configuration

Tue, 27 Jul 2010 11:05:20 +0000

[...] Of Oracle Data Warehousing – Balanced Hardware Configuration December 22nd, 2009[back to Introduction]If you want to build a house that will stand the test of time, you need to build on a solid [...]

By: The Core Performance Fundamentals Of Oracle Data Warehousing – Set Processing vs Row Processing

Tue, 20 Jul 2010 21:01:54 +0000

[...] – Set Processing vs Row Processing July 20th, 2010 Goto comments Leave a comment[back to Introduction]In over six years of doing data warehouse POCs and benchmarks for clients there is one area that I [...]

By: The Core Performance Fundamentals Of Oracle Data Warehousing – Set Processing vs Row Processing

Tue, 20 Jul 2010 21:01:54 +0000

By: DW Introduction « Oracle Scratchpad

DW Introduction « Oracle Scratchpad — Mon, 26 Apr 2010 16:29:20 +0000

[...] series on “Core Performance Fundamentals of Oracle Data Warehousing”. Here’s his catalogue of the first four or five articles in the [...]

By: The Core Performance Fundamentals Of Oracle Data Warehousing – Data Loading

Fri, 23 Apr 2010 18:37:46 +0000

[...] Oracle Data Warehousing – Data Loading April 23rd, 2010 Goto comments Leave a comment[back to Introduction]Getting flat file data into your Oracle data warehouse is likely a daily (or more [...]

By: Greg Rahn

Greg Rahn — Fri, 08 Jan 2010 19:30:59 +0000

@Eric Sun

This comment doesn’t have much of anything to do with the post and this is not a support forum but I will give you some advice this one time. Parallel Execution (PX) will use as few nodes as possible to satisfy the execution because it is more efficient to do so (why run something on 4 computers when there are enough resources to run it on 1?). Given you have only 8 slaves (maybe that is DOP=8 and there are 8 producer slaves and 8 consumer slaves, or you have DOP=4 and 4 produces/4 consumers) and 4 nodes, it seems likely that in either case 1 single node can satisfy the PX requirements. If you want to use all 4 nodes, use a DOP of the number of CPUs in the cluster (or 2x the CPUs [cpu_count] in the cluster, which is the “default” DOP).

By: Eric Sun

Eric Sun — Fri, 08 Jan 2010 06:14:58 +0000

Hi Greg,

Do you which parameters in Oracle 11g and 10g determine how the coordinator/slave queries are distributed among nodes? I’ve observed a strange thing in my 4-node RAC. The coordinator is on node-A, but all the 8 slaves are all on node-B. The slaves are never on the same node as coordinator. Even for full partition-wise joins among hash-partitioned tables, the slaves are never distributed across multiple nodes.

Load balance is enabled in the TNS entry. Is there any fundamental setting that I missed?

Thank you!

By: Greg Rahn

Greg Rahn — Mon, 04 Jan 2010 01:18:25 +0000

@Marco Gralike
I spend 99% of my time in the Data Warehouse area so the topics are more DW focused, but glad that Doug guy invited you. ;)

By: Marco Gralike

Marco Gralike — Thu, 24 Dec 2009 17:03:56 +0000

Hi Greg,

indeed, I enjoyed the “chalk and talk” session very much, one of my highlights during OOW and grateful that that Scottish guy tagged me along (toys included…). I really think your insights are applicable on more than Data Warehousing so one happy guy will follow the series and see how I can make use of them in my “domain”…

;-)

By: Greg Rahn

Greg Rahn — Wed, 16 Dec 2009 05:55:33 +0000

@Ram

Thanks for the suggestion. I’ll try and work that in.