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 stutiawasthi@hcl.com designates
 203.105.185.23 as permitted sender)
From: Stuti Awasthi <stutiawasthi@hcl.com>
To: "user@hbase.apache.org" <user@hbase.apache.org>
Date: Tue, 30 Aug 2011 09:20:42 +0530
Subject: RE: Scan not working properly on composite keys
Thread-Topic: Scan not working properly on composite keys
Thread-Index: AcxmZOPOy150xh7pRDGx8gtUsHPhhwAYwTmg
Message-ID: 
 <7D9AF4B98807C54EBADEA75DF6D5ACB7951BE8@NDA-HCLC-EVS04.HCLC.CORP.HCL.IN>
References: 
 <7D9AF4B98807C54EBADEA75DF6D5ACB7951959@NDA-HCLC-EVS04.HCLC.CORP.HCL.IN>
 <2D6136772A13B84E95DF6DA79E85A9F00158E398C204@NSPEXMBX-A.the-lab.llnl.gov>
 <7D9AF4B98807C54EBADEA75DF6D5ACB7951B7B@NDA-HCLC-EVS04.HCLC.CORP.HCL.IN>
 <A97E620A-65C8-4DF0-B526-F88FC339F12D@salesforce.com>
In-Reply-To: <A97E620A-65C8-4DF0-B526-F88FC339F12D@salesforce.com>
Accept-Language: en-US
Content-Language: en-US
acceptlanguage: en-US
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0

Hi Ian,

Thanks for explaining this. :)

-----Original Message-----
From: Ian Varley [mailto:ivarley@salesforce.com]=20
Sent: Monday, August 29, 2011 9:31 PM
To: user@hbase.apache.org
Subject: Re: Scan not working properly on composite keys

Stuti,

The rows are physically sorted on disk (and in memory) according to the row=
 keys you define, and that's the only way that HBase can access them (unlik=
e a relational database, it doesn't have built in indexes that allow you to=
 access rows by something other than their physical sort order). So having =
a general wildcard lookup feature wouldn't be practical; in reality, it wou=
ld have to scan ever row in the table to see which ones match your pattern.=
=20

That said, you could build a little wrapper on top of your access code as f=
ollows (pseudocode):

function getRowsByUser(string user) {
	prefixes =3D "admin", "viewer", "etc";
	for each prefix in prefixes {
		result.add(getRowsFromHBase(prefix + user));
	}
	return result;
}

That way, at least your scan is linear in the number of prefixes--it just d=
oes one constant-time seek to the start of the rows for the given user, for=
 each of the prefixes you know about.

You can't expect functionality like this to be built in to HBase, because t=
o HBase, row keys are simply byte arrays, nothing more. It does nothing to =
enforce structure on them, that's entirely your application's job.

Ian

On Aug 29, 2011, at 10:49 AM, Stuti Awasthi wrote:

> Hi,
>=20
> Sure, I tried also by prefixing with admin, and it was working fine. I wa=
s just wondering that is there any way that we can get the information from=
 Hbase using some part of Composite keys or wild card characters ?
>=20
> Thanks
>=20
> -----Original Message-----
> From: Buttler, David [mailto:buttler1@llnl.gov]=20
> Sent: Monday, August 29, 2011 8:46 PM
> To: user@hbase.apache.org
> Subject: RE: Scan not working properly on composite keys
>=20
> The scan started at "viewer/user3" because that is the next key after "us=
er2".  If your keys have a prefix, like admin or viewer, then you need to p=
repend that to your start row as well.  Hbase has no magic where it can int=
uit your intent.
>=20
> Dave
>=20
> -----Original Message-----
> From: Stuti Awasthi [mailto:stutiawasthi@hcl.com]
> Sent: Monday, August 29, 2011 12:39 AM
> To: user@hbase.apache.org
> Subject: Scan not working properly on composite keys
>=20
> Hi Friends,
>=20
> I have a query regarding scan operation on composite keys.
>=20
> I have table as follows :
>=20
> admin/user1                                       column=3Dinfo:email, ti=
mestamp=3D1314594930540, value=3Dabc
> admin/ user1                                    column=3Dinfo:mobile, tim=
estamp=3D1314594940686, value=3D123456
> admin/ user1                                    column=3Dinfo:name, times=
tamp=3D1314594915091, value=3Ddef
> admin/user2                                     column=3Dinfo:email, time=
stamp=3D1314597627865, value=3Ddef
> admin/ user2                                    column=3Dinfo:mobile, tim=
estamp=3D1314597636663, value=3D67890
> admin/ user2                                    column=3Dinfo:name, times=
tamp=3D1314597614585, value=3Dpqr
> viewer/user3                                     column=3Dinfo:email, tim=
estamp=3D1314601194373, value=3Dpqr
> viewer/ user3                                    column=3Dinfo:mobile, ti=
mestamp=3D1314601256479, value=3D12345
> viewer/ user3                                   column=3Dinfo:name, times=
tamp=3D1314601207662, value=3Dxyz
>=20
> I want to find all the records of "user2". For this I tried following que=
ry but it is giving me all records of user3 :
>=20
> hbase(main):005:0> scan 'users', {STARTROW=3D>'user2'}
> ROW                                         COLUMN+CELL
> viewer/user3                          column=3Dinfo:email, timestamp=3D13=
14601194373, value=3Dsatyendrap@hcl.com
> viewer/ user3                          column=3Dinfo:mobile, timestamp=3D=
1314601256479, value=3D37498734893
> viewer/ user3                          column=3Dinfo:name, timestamp=3D13=
14601207662, value=3Dsatyendra pandey
> 1 row(s) in 0.0330 seconds
>=20
> I also tried various queries for this but not getting correct result. I a=
m definitely missing something. Please help
>=20
> Thanks & Regards
> Stuti Awasthi
> Sr Specialist
>=20
>=20
> ________________________________
> ::DISCLAIMER::
> -------------------------------------------------------------------------=
----------------------------------------------
>=20
> The contents of this e-mail and any attachment(s) are confidential and in=
tended for the named recipient(s) only.
> It shall not attach any liability on the originator or HCL or its affilia=
tes. Any views or opinions presented in this email are solely those of the =
author and may not necessarily reflect the opinions of HCL or its affiliate=
s.
> Any form of reproduction, dissemination, copying, disclosure, modificatio=
n, distribution and / or publication of this message without the prior writ=
ten consent of the author of this e-mail is strictly prohibited. If you hav=
e received this email in error please delete it and notify the sender immed=
iately. Before opening any mail and attachments please check them for virus=
es and defect.
>=20
> -------------------------------------------------------------------------=
----------------------------------------------