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 linlindu2009@hotmail.com
 designates 65.54.190.89 as permitted sender)
Message-ID: <BAY179-W17E247391B699BD663FAAD5760@phx.gbl>
Content-Type: multipart/alternative;
	boundary="_13347522-55e5-4339-a687-8f88c5202258_"
From: Linlin Du <linlindu2009@hotmail.com>
To: "user@hbase.apache.org" <user@hbase.apache.org>
Subject: about region merging
Date: Wed, 12 Mar 2014 10:40:50 -0300
Importance: Normal
MIME-Version: 1.0

--_13347522-55e5-4339-a687-8f88c5202258_
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Hi all=2C

Our usage of hbase is not very right and in many cases=2C time stamp is use=
d as the prefix
of the keys. And it is too late to get that fixed now.
As more data comes in=2C more new regions will be spawned from the existing=
 regions. And after
the retention period expires=2C many regions are almost empty. It is desira=
ble to have these
empty region deleted. Here are my questions:

1 If a timestamp is used in the key (maybe not as prefix=2C e.g.=2C in salt=
ing) and there is a retention period=2C
after a region's data expires=2C the region becomes empty and is never used=
 again. Is that right?

2 How to find out when a region was last modified/read?

3 Is there any easy way to merge small regions into a big one?

Thanks.

Linlin
 		 	   		  =

--_13347522-55e5-4339-a687-8f88c5202258_--