Mailing-List: contact user-help@hbase.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@hbase.apache.org
Received-SPF: pass (athena.apache.org: domain of izquierdo@strands.com
 designates 217.116.18.226 as permitted sender)
Subject: Re: Thousands of tablesq
From: =?ISO-8859-1?Q?H=E9ctor?= Izquierdo Seliva <izquierdo@strands.com>
To: user@hbase.apache.org
In-Reply-To: <AANLkTinr3+wJsjoEW_wThDdXRGjsus3ySbRmZxB1vqFp@mail.gmail.com>
References: <1280501095.5980.3.camel@mierdi-laptop>
	 <AANLkTimARyAfoTvLL8R=W1ksNEncCFXisNPLDztMUxjG@mail.gmail.com>
	 <1280507437.5980.10.camel@mierdi-laptop>
	 <AANLkTinr3+wJsjoEW_wThDdXRGjsus3ySbRmZxB1vqFp@mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"
Date: Mon, 02 Aug 2010 12:14:45 +0200
Message-ID: <1280744085.5980.15.camel@mierdi-laptop>
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit

> If you expect that some contiguous rows would be really overused, then
> change the row key. UUIDs for example would spread them all over the
> regions.
> 
> In 0.20 you can do a close_region in the shell, that will move the
> region to the first region servers that checks. In 0.90 we are working
> on better load balancing, more properly tuned to region traffic.

I'm on 0.89. Is any of this work in there? I can not use UUIDs, as I
need to address rows by key, but I could add a hash of the key to the
begining so keys are more evenly distributed. This is of course in case
of going to a single table.

I filled a 3 node cluster with data (around 6 GB), and the read
performance was very bad (lots of LRU flushes and misses). I'll test
this approach and see if it works better. 

Thank you very much!