Re: deleting many rows from a table

From: UXDBA <unixdba73_at_googlemail.com>
Date: Thu, 14 Jan 2010 05:06:48 -0800 (PST)
Message-ID: <e393de28-de5f-4802-ab5c-cbc7978ca9e1_at_e27g2000yqd.googlegroups.com>



Thanks Gints .

But Live table T will have ongoing transaction.

So step#2 would also require downtime?

On Jan 14, 12:53 pm, Gints Plivna <gints.pli..._at_gmail.com> wrote:
> How about:
> 1) CREATE table t1 with the same structure and necessary storage
> definitions
> 2) INSERT /*+ append */ INTO t1 select only necessary rows
> 3) drop old table t
> 4) RENAME t1 to t;
>
> Of course it means, that all privileges should be regranted and
> dependant procedural units recompiled.
>
> Minimum downtime (only steps 3 and 4) - the only problem is need for
> extra space and recompile units/regrant privileges.
>
> Gints Plivnahttp://www.gplivna.eu

Thanks Received on Thu Jan 14 2010 - 07:06:48 CST

Original text of this message