Date: Fri, 11 Jan 2008 23:43:58 +0100 (CET)
From: Pomarede Nicolas <npomarede@corp.free.fr>
To: Andrew Sullivan <ajs@crankycanuck.ca>
cc: pgsql-performance@postgresql.org
Subject: Re: Best way to index IP data?
In-Reply-To: <20080111222403.GA7527@crankycanuck.ca>
Message-ID: <Pine.LNX.4.64.0801112336500.30027@localhost>
References: <200801101614.54729.kevin@kevinkempterllc.com>
	<DA6DE53A-2F69-46F3-8C96-B341ADCDDFAF@blighty.com>
	<36e682920801101541o7626bbddj428617144240c554@mail.gmail.com>
	<Pine.LNX.4.64.0801111025390.30762@localhost>
	<13179.1200064791@sss.pgh.pa.us>
	<20080111162747.GT5294@mathom.us> <24044.1200082058@sss.pgh.pa.us>
	<20080111212703.GU5294@mathom.us>
	<20080111213710.GD8260@alvh.no-ip.org>
	<20080111220235.GX5294@mathom.us>
	<20080111222403.GA7527@crankycanuck.ca>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed

On Fri, 11 Jan 2008, Andrew Sullivan wrote:

> On Fri, Jan 11, 2008 at 05:02:36PM -0500, Michael Stone wrote:
>
>> networks), but there's a conspicuous lack of a type for (hosts). I
>> suppose if you really are sure that you want to store hosts and not
>> networks
>
> Well, part of the trouble is that in the CIDR world, an IP without a netmask
> can be dangerously ambiguous.  I can see why the design is as it is for that
> reason.  (But I understand the problem.)
>
> A
>

Yes, in fact it all depends on the meaning you give to an IP.

If you want to store subnets, then you need an IP and a netmask, but if 
you just want to store the IP of a particular equipment (that is, the IP 
that will be refered to in the TCP/IP header), then there's no ambiguity, 
you just need 4 bytes to describe this IP.

And it's true for IPv6 too, storing an IP that refer to an end point and 
not a subnet is requiring twice as much data as needed, because the 
netmask would always be ff:ff:ff:..:ff

So, for people dealing with large database of IPs, it would be nice to be 
able to save 50% of the corresponding disk/cache/ram space for these IPs.


Nicolas