Mailing-List: contact user-help@hbase.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@hbase.apache.org
Received-SPF: pass (nike.apache.org: domain of kiran.sarvabhotla@gmail.com
 designates 209.85.192.173 as permitted sender)
MIME-Version: 1.0
Date: Tue, 3 Sep 2013 15:58:57 +0530
Message-ID: 
 <CAMBJfV=2897EFRqtcNsOmc9e3Ei7HfC3UNp75uOOnq-y7XRmow@mail.gmail.com>
Subject: Performing intra row scans
From: kiran <kiran.sarvabhotla@gmail.com>
To: user@hbase.apache.org
Content-Type: multipart/alternative; boundary=047d7b2e10430e2fda04e5782825

--047d7b2e10430e2fda04e5782825
Content-Type: text/plain; charset=ISO-8859-1

Dear All,

We have a row where there are a million entries inserted as columns per
day. The qualifier structure "prefix-reversetimestamp".
I am performing a day wise slice of the columns using Columnrangefilter.
When I tried to fetch columns that span within 3 days from the current
time. There was no problem at all.

When, I tried to increase the number of days to morethan 3, then my entire
cluster stops accepting new requests.

Is it the case of region hot swapping due to my schema design?

Is it a good to have million columns under a single row and growing each
day.

It is a production cluster, so I can't afford this outage.

Please suggest me the workaround to handle this.

Hbase version: 0.94.1

My sample code:
Scan scan = new Scan(rowBytes, rowBytes);
scan.setBatch(200);
byte[] startPrefix = prefix + "-" + (Long.Max_value -
currentimeroundedtoday);
byte[] endPrefix  = prefix + "-" + (Long.Max_value -
(currenttimeroundedtoday - 7 days));
ColumnRangeFilger crf = new ColumnRangeFilter(startPrefix, true, endPrefix,
true);
scan.setFilter(crf);


-- 
Thank you
Kiran Sarvabhotla

-----Even a correct decision is wrong when it is taken late

--047d7b2e10430e2fda04e5782825--