Re: Bloom filters

From: joel garry <joel-garry_at_home.com>
Date: Mon, 10 Dec 2012 08:34:03 -0800 (PST)
Message-ID: <8bbde826-aa6a-42df-812a-5001309e4a18_at_g7g2000pbi.googlegroups.com>

On Dec 9, 11:53 am, Mladen Gogala <gogala.mla..._at_gmail.com> wrote:

...
> Two hints can be used: px_join_filter and no_px_join_filter; and there
> are a couple of views for monitoring Bloom filters:
>
> - v$sql_join_filter: information about Bloom filters; number of rows
> filtered out and probed by Bloom filters
> - v$pq_tqstat: check reduced communication due to usage of Bloom filters
> **************************************************************************
>
> The author is an Oracle employee, so I searched the Metalink. The search
> has produced the document 912330.1 which, while less detailed than the
> post above still confirms it. Of course, as with all new features, there
> are "features":
>
> Bug 12637294: BLOOM PARTITION PRUNING MAY CAUSE DEADLOCK ORA-60
>
> Funny thing about this particular "feature" is that it is marked as fixed
> in release 12.1.

Those "fixed in" statements can be misleading, you have to look for backports. You can get the 11.2.0.3 patch for exadata, or, can I believe my eyes, hp-ux Itanium.

jg

--
_at_home.com is bogus.
http://video-games.pricegrabber.com/cdos+readme+first/results.html/

Received on Mon Dec 10 2012 - 17:34:03 CET