Re: Wich hardware suits best for large full-text indexed

Dann Corbit wrote:

>I see that only table documentos has a unique index. Is it really so
>that none of the other tables has a key to uniquely identify a record?
>Perhaps the others have a unique attribute but it was never declared
>when forming the index?
>
>
I thougth that a primary key would create a unique index implictly.

>I do not remember which version of PostgreSQL you are using, but there
>is now a CLUSTER command. Now, I am speaking with my experience in
>other database systems, and so perhaps the PostgreSQL group may find it
>necessary to correct me. I have only recently began working with
>PostgreSQL versions beyond 7.1.3 which had no cluster command (or at
>least I was not aware of one). In other database systems, a clustered
>index is incredibly valuable. In fact, a unique, clustered index is
>always a silver bullet to marvelous performance. Perhaps some of the
>experts can give measurements as to the effectiveness of clustering in
>PostgreSQL. Here is the link for the cluster documentation:
>http://www.postgresql.org/docs/current/static/sql-cluster.html
>
>>From the description, clustering is expensive and your tables are large.
>So it should be an infrequent operation.
>
>
Sure, my pgsql is version 7.3.4 (I was obviosly mistaken when I wrote
7.4.3 in a previous email) I used to do clustering in some tables, but
is a expensive operation as said...

--
Diogo Biazus
diogo(at)ikono(dot)com(dot)br
http://www.ikono.com.br