Hari Om wrote:
>
> This message is for "Connor McDonald" with reference to his reply @
> http://groups.google.co.uk/groups?dq=&hl=en&lr=lang_en&ie=UTF-8&selm=3CC4534C.5DA8%40yahoo.com&rnum=40
>
> I have a question about it. In your reply you gave following example:
> ----------------------------------------------------------------------------
> create table BOEING1 ( x number, y char(100));
> create table BOEING2 ( x number, y char(100));
> create index BOEING1 on BOEING1 ( x);
> create index BOEING2IX on BOEING2IX ( x);
>
> 10000 rows inserted into each table, 200 rows for each value of x from
> 0
> to 49. So after running an analyze...
>
> analyze table BOEING1 compute statistics;
> analyze table BOEING2 compute statistics;
> etc...
> ----------------------------------------------------------------------------
>
> what is BOEING2IX in the create index command....?is it a typo....?
> should it be
> BOEING2 instead? Also, what are you trying to show with this
> example....? It seems that I am lost in your example....could you
> please explain your point again.
>
> THANKS a MILLIONS!
>
> HARI OM
The point I am making is that if you need to obtain (say) 20% of the
rows from a table, you cannot from this information alone decide whether
you need to use an index or a full scan. What really matters is the
data distribution not what proportion of rows you need to get back from
your query
and yes, there was a typo in my scripts way back then :-)
Received on Fri Sep 19 2003 - 07:03:44 CDT
