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 lists@nabble.com designates
 216.139.236.26 as permitted sender)
Message-ID: <33128226.post@talk.nabble.com>
Date: Thu, 12 Jan 2012 06:50:28 -0800 (PST)
From: fullysane <fullysane@msn.com>
To: hbase-user@hadoop.apache.org
Subject: Re: Question about HBase for OLTP
In-Reply-To: <33107782.post@talk.nabble.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
References: <33107782.post@talk.nabble.com>


Hi Mike:

Thereason I am thinking Hbase for OLTP is that I need a column-based
(key-value pair)OLTP DBMS which allows me not to predefine columns for a
table and can add new column to a table on the fly like Hbase does. Alougth
this can be done in any RDBMS with so called skinny and tall table contain
only key and value columns and then pivoting the data to look like a
traditionla spead sheet like table, but the pivoting perforamnce will be a
drag for query.

Any idea to solve this problem?

Thanks,


fullysane wrote:
>=20
> Hi
>=20
> I know HBase is designed for OLAP, query intensive type of applications.
> But I like the flexibility feature of its column-base architecture which
> allows me having no need to predefine every column of a table and I can
> dynamically add new column with value in my OLTP application code and
> capture its meta data information.=20
>=20
> My question is basically about if we can use HBase for OLTP application
> database. I know Hbase works well with Inserting column data of a row key
> and set new version for the new piece of the data, and not so well for
> updating and deleting existing piece of data. However, if I turn OLTP
> update and delete operations into all insertion of new version of colum
> data as I described below:=20
> For OLTP data update, if I set my table column family=E2=80=99s versionin=
g to 1
> and always do insert (put) when there is need to update an existing data
> row columns, and let Hbase to handle the delete of the old versions
> through DB garbage collection.
> For OLTP data delete, I can use inserting new version on a flag field to
> =E2=80=9Cdeleted=E2=80=9D, which is a logical delete, and have some batch=
 job to clean up
> all logically deleted rows later.=20
>=20
> Will the above scenario work for using HBase for an OLTP application? Any
> flaws on doing it?
>=20
> Can some one share the experiences of using HBase for OLTP applications?
>=20
> Thanks,
>=20
>=20

--=20
View this message in context: http://old.nabble.com/Question-about-HBase-fo=
r-OLTP-tp33107782p33128226.html
Sent from the HBase User mailing list archive at Nabble.com.