Forwarded Message
Date: Fri, 20 Apr 2001 08:37:22 -0700 (PDT)
Message-Id: <[EMAIL PROTECTED]>
To: Paul A Vixie <[EMAIL PROTECTED]>
Subject: Re: Appliance caching server configuration database schema
In-Reply-To: <[EMAIL PROTECTED]>
From: [EMAIL PROTECTED] (Andreas Gustafsson)
Paul A.
> AFAIK we never actually *had* IPV6 support in those datatypes, only
> stubs for it.
the inet_net_pton implementation that was brought in from BIND had its
IPv6 portions scrubbed. micro-over-optimization of the contributed
"bitncmp" caused the "ipv4 as int" assumption to reoccur. i'm going to
> See bytea, though its presentation format leaves something to be desired IMHO
>
> > how would someone be expected to store, say, a GIF image in a TOAST text?
>
> One would not. A TOASTed bytea is the appropriate column type.
thanks -- that's EXACTLY what i needed.
for my pgcat utility i now know i have to use \nnn octal quoting for
nonprintables in the generated INSERT commands. but in testing, i
found the following oddity. this is in 7.1-b1 (cvs-current).
vixie=> create table foo ( bar text );
CREATE
vixie=> insert into foo value
> > anybody know what i could be doing wrong? (i'm also wondering why SELECT
> > takes ~250ms whereas INSERT takes ~70ms... seems counterintuitive, unless
> > TOAST is doing a LOT better than i think.)
>
> I would think that this is entirely due to planning the query. An INSERT
> has no decision
the prior results (http://www.vix.com/~vixie/pgsql-results.png) showed:
~70ms usual INSERT time (~1.5sec -> ~1.25sec occasional)
~250ms usual SELECT time (~1.5sec occasional)
changing the attribute i key by to be PRIMARY KEY improved things a lot;
the new results (http://www.vix.
> 1) Have you ran vacuum analyze after all these inserts to update database
> statistics? :) Without vacuum, pgsql will opt to table scan even when
> there's an index.
i hadn't, but i did, and it didn't make that particular difference:
vixie=# explain select file from rss where addr = '1
(plz cc me on your replies, i'm not on pgsql-hackers for some reason.)
http://www.vix.com/~vixie/results-psql.png shows a gnuplot of the wall time
of 70K executions of "pgcat" (shown below) using a CIDR key and TEXT value.
(this is for storing the MAPS RSS, which we presently have in flat files.)
