Home » Other » General » Statistical Analysis of Oracle Table Clustering (11.1.0.7.0)
icon14.gif  Statistical Analysis of Oracle Table Clustering [message #452593] Thu, 22 April 2010 22:58 Go to next message
rahulvb
Messages: 924
Registered: October 2009
Location: Somewhere Near Equator.
Senior Member
Hi All,

I have came across very nice Article so sharing it.

Moderator if this post does not belongs to this group please help me to to move it.

Statistical Analysis of Oracle Table Clustering

Happy Reading.
Re: Statistical Analysis of Oracle Table Clustering [message #452683 is a reply to message #452593] Fri, 23 April 2010 06:15 Go to previous messageGo to next message
Mahesh Rajendran
Messages: 10707
Registered: March 2002
Location: oracleDocoVille
Senior Member
Account Moderator
It is talking about Oracle 8i which is a decade old now.
Re: Statistical Analysis of Oracle Table Clustering [message #452776 is a reply to message #452683] Sat, 24 April 2010 02:48 Go to previous messageGo to next message
rahulvb
Messages: 924
Registered: October 2009
Location: Somewhere Near Equator.
Senior Member
Mahesh Rajendran wrote on Fri, 23 April 2010 06:15
It is talking about Oracle 8i which is a decade old now.


But Fundamentals should remain Same !! Right?
Re: Statistical Analysis of Oracle Table Clustering [message #452791 is a reply to message #452683] Sat, 24 April 2010 10:12 Go to previous messageGo to next message
rleishman
Messages: 3728
Registered: October 2005
Location: Melbourne, Australia
Senior Member
Interesting, perhaps. Not sure if it is useful though.

What is it giving you? Seems like it provides an arbitrary measure (degree of chaining) of the optimality of a cluster. By trying different cluster settings, you can derive a predicted chaining probability for each configuration.

And higher probability of chaining is bad, right? Hang on, if you increase the density of your cluster, you increase the chaining count, but you also reduce the size of the table, which is good for multi-block range scans and full table scans.

So this chaining probability; we can measure it, but it still gives no way to tell what is "good" and what is "bad".

But that's not the worst of it. This article assumes you have already decided to use a cluster. But how do you make that decision? What are the alternatives? How do they compare with a cluster for different types of access?

I've done a small amount of research on clusters and I've had very inconclusive results. I have found for a straight join of two tables, a partition-wise hash join is faster than a cluster join. This tells me that clusters are sub-optimal for data warehouses that perform large joins.

I suspect (without proof) that the optimal use of clusters is in small volume lookups of related rows in a high-concurrency environment (many simultaneous users).

Ross Leishman
Re: Statistical Analysis of Oracle Table Clustering [message #452875 is a reply to message #452791] Sun, 25 April 2010 21:24 Go to previous message
rahulvb
Messages: 924
Registered: October 2009
Location: Somewhere Near Equator.
Senior Member
rleishman wrote on Sat, 24 April 2010 10:12
This article assumes you have already decided to use a cluster.
Ross Leishman



Yes,even I think so that Article was written with prejudice mindset.
But overall nice Article.
8/10 .
Previous Topic: why oracle
Next Topic: Oracle Partner Network
Goto Forum:
  


Current Time: Thu Mar 28 07:55:09 CDT 2024