Mailing-List: contact dev-help@couchdb.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@couchdb.apache.org
Received-SPF: pass (nike.apache.org: domain of gcdcd-couchdb-2@m.gmane.org
 designates 80.91.229.12 as permitted sender)
To: dev@couchdb.apache.org
From: Roger Binns <rogerb@rogerbinns.com>
Subject: Re: Objective criteria
Date: Wed, 13 Jan 2010 23:30:04 -0800
Lines: 45
Message-ID: <himh67$oci$1@ger.gmane.org>
References: <hilje3$gf1$1@ger.gmane.org>
	 <e282921e1001131514q101e399t9704f7892983c894@mail.gmail.com>
	 <hillfc$mih$1@ger.gmane.org>
	 <661B0EA8-4CE8-435A-8175-6A4C3FBEA5E9@apache.org>
	 <him2jd$me1$1@ger.gmane.org>
 <e282921e1001132046o593f3e57g67caaca46430072@mail.gmail.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
User-Agent: Thunderbird 2.0.0.23 (X11/20090817)
In-Reply-To: <e282921e1001132046o593f3e57g67caaca46430072@mail.gmail.com>
Sender: news <news@ger.gmane.org>

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Chris Anderson wrote:
>> I know your data is on the large side and CouchDB doesn't auto-cluster

Ah, the clue!  I don't think my data is "large" by any measure (10 million
docs, 2GB of JSON).  SQLite (note *lite* in the name) doesn't break a sweat.
 It only occupies 40% of a DVD.  Picking a random low end machine from Dell
shows that they ship with a minimum of 1GB of RAM and ideally want you to
buy 2GB.  Something that fits in the RAM of a $350 machine from Dell is not
what I would consider large!  The data fits in my machine's RAM 4 times
over.  Can you even buy USB sticks or SD cards these days smaller than 2GB?
 You could fit 15 copies of my data and an operating system in the smallest
SSD drives.

GMail's initial quota however many years ago was 1GB.  Keith Packard's email
is half a million messages but 5GB of data - http://keithp.com/blogs/notmuch/

My machine has 350,000 files and directories (excluding backups which
duplicate many of those multiple times over).  This is a similar order of
magnitude as my data set (and several times larger if counting backups).
(Note I am just talking about if you constructed a database of file and
directory names, information about them etc - not the contents.)

My deployment plans are the opposite of clustering and partitioning as my
data set is so small!  I wanted to put a copy of CouchDB on each and every
server and have them replicated to each other rather than dedicated
networked data servers.

If this kind of (trivial!) size means clustering, paritioning etc then
CouchDB is not remotely appropriate for my circumstances, and probably not
for people recording file and email databases.  I only wish there was
documentation somewhere saying what "normal" sizes are for CouchDB and
expectations for them.

Roger
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAktOx/wACgkQmOOfHg372QS8RgCgp5/GTZCHyZG3Sf8qaZMAAppe
oM8AniZXLa2tXOw78w/N29shsQqOMBWT
=46bm
-----END PGP SIGNATURE-----