Re: Challenge: Partitioning is a wrong idea

From: <fitzjarrell_at_cox.net>
Date: 7 Apr 2005 15:12:58 -0700
Message-ID: <1112911978.705661.271110@l41g2000cwc.googlegroups.com>

mikharakiri_nosp..._at_yahoo.com wrote:

> hpuxrac wrote:
> > Have you read Jonathon Lewis's book?  He has about 50 pages I think
> > about partitioning.
> >
> > Also Tom Kyte's effective oracle by design.  Chapter 1 "all of it"
> plus
> > at least the section on partitioning or not?
> >
> > You need to understand a bunch more about the subject ( my opinion

).

> >
> > Remember "no silver bullets" just a lot of techniques that can be
> very
> > useful if you understand the pro's and con's, when to use them,

when

> > not to, how to test whether they are useful for your environment or
> > not.
> >
> > "It depends" ... but you need to do a bunch more legwork.
>
> Please don't "no silver bullets" bu***it me. If I ask the same

question
> about bitmapped indexes, or marerialized views, I would be immediately
> referred to some research and whitepapers. I hold nothing against both
> books -- which are indeed excellent books for the right audience. Yet,
> I expect you to provide more substantial argument.

And it's expected of you to realise you're simply trolling; we do. Had you any real knowledge of the concept of partitioning such a statement as your original post would not have been issued. Let's look at your original 'troll' again. You stated:

| I'm returning to this topic because of rumors that "interval
| partitiong" would be the next new shiny administrative toy.
|
| Some time ago I asked what are the benefits of partitioned table
| compared to partitioned view. Now I ask why partition a table *at
all*?
| Is there logical difference between large and small table? Next, why
| partition by a certain column and not the other one? What range to
| choose? OK, monkey type DBA is not supposed to ask this kind of
| questions, but the idea of "scientific method" popularised by
Jonathan
| and the others recently certainly legitimaze it.
|
| It is undeniable that partitioning concept introduces extra
| complications. You have to be aware of many extra technicalities:
what
| is partition prunning, what is partition wise join, etc.
|
| Returning to the "scientific method" theme, a single test case can
| settle the issue once and forever. Create large table, fill in with
the
| data, and show how much does it take to accomplish a certain
| administrative tasks in both cases. I claim that for any
administrative
| task you suggest, I would find a way to accomplish it in reasonable
| time with normal table. In other words: partitioning buys up
*nothing*.
| Any takers?

If you are thoroughly convinced of that thought then why post to the newsgroup knowing you'll discount and disregard any and all rebuttals? You've certainly shown such behaviour thus far. Be that as it may, your contention is:

| I claim that for any administrative
| task you suggest, I would find a way to accomplish it in reasonable
| time with normal table. In other words: partitioning buys up
*nothing*.

It appears the onus is upon you to provide your assessment of 'reasonable time', as what may be reasonable to you may be unreasonable to others in your position. The first example, of deleting a years' worth of data out of a 1 terabyte table containing 8 years of records is an excellent choice, and you summarily discounted the response, as should be expected from someone trolling a newsgroup to fan flames. A simple alter table ... drop partition ... would be far faster and generate much less redo and much less rollback/undo than a delete from sales_history where txn_date <= ..., and would consume less cpu in the process. Yet you persist in thinking in a box, a small box, it would appear, and refuse to accept any advice or response contrary to your pre-conceived (or is that ill-conceived) notions. To be honest, I respect Thomas Kyte and Jonathan Lewis far more than I could ever respect you or your 'knowledge', and, as I've successfully used their guidelines and recommendations on production systems with regard to partitioned tables I can only take your verbiage as the aimless ramblings of someone with the intention of causing trouble. You've been directed to read the texts by these two individuals, yet you claim these texts have nothing to do with YOUR trolling nonsense. I, as well as others in this newsgroup, disagree.

You've received several thoughtful and meaningful responses to your noise, yet you make more rude comments claiming they are essentially idiotic submissions unworthy of your consideration. Hans Forbich has offered two such responses, both well-thought and on-target. That you refuse to admit their value certainly bolsters the general opinion that you are attempting to disrupt this newsgroup with your drivel. It is time you carefully examine the responses again, and realise their worth. Of course, you'll state they are worth nothing, as they don't answer your 'burning question' on the value of partitioning. You've tried this once before, as you readily admit, with your troll on partitioned tables versus partitioned views. That obviously was not enough to satisfy your need to troll, so you again post arrogant assertions in hopes of a flame war. It is too bad we won't aquiesce to your request.

You'll discount this respones, as well, as it didn't offer any 'scientific proof' you so adamantly demand. The truth is table partitioning is not a cut-and-dried scientific operation, it's a subjective task best left to those who understand its nature. Partitioning is not for everyone, nor for every situation. Don't try to disuade people simiply because your closed mind can't grasp the benefits partiioning can provide.

David Fitzjarrell Received on Thu Apr 07 2005 - 17:12:58 CDT