You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@hbase.apache.org by Nishant Khurana <ni...@gmail.com> on 2008/11/22 06:27:52 UTC
Question of Column families in Hbase
Hi,
I was looking at the API Docs and HQL shell to create tables for Hbase. I
couldn't figure out :
1) How to create column families such that each column family contains more
than one column through shell.
2) Also if I want to do it programmatically, HColumnDescriptor class only
lets you create column families with single columns. Is there a way to do it
?
Also If I am not wrong, we can store multiple values of each column in each
column family ?
Is there a way I can do a bulk import from a structured text file into Hbase
through the shell or any script already available ?
Thanks
--
Nishant Khurana
Candidate for Masters in Engineering (Dec 2009)
Computer and Information Science
School of Engineering and Applied Science
University of Pennsylvania
Re: Question of Column families in Hbase
Posted by stack <st...@duboce.net>.
Nishant Khurana wrote:
> Hi,
> Thanks for pointing me to the new link. A google search on Hbase Shell was
> returning the old one at the top so was trying to use that.
I've capitalized and tripled the banner at the top of that page that
tries to say this old shell no longer applies.
> Sp from your
> answers below, I think we always have to define column families in the
> schema first and then alter table to add fully qualified column names to add
> multiple columns to a column families.
>
No need to do table 'alter'. Just add new columns when you do an
update. It'll just work as long as the column has an existing column
family's prefix.
St.Ack
> I will try to use the scripts as mentioned there.
> Thanks
>
> On Sat, Nov 22, 2008 at 1:47 AM, Michael Stack <st...@duboce.net> wrote:
>
>
>> Nishant Khurana wrote:
>>
>>
>>> Hi,
>>> I was looking at the API Docs and HQL shell to create tables for Hbase. I
>>> couldn't figure out :
>>> 1) How to create column families such that each column family contains
>>> more
>>> than one column through shell.
>>>
>>>
>>>
>> Please tell us how you tripped over HQL. It was remove from hbase quite a
>> while ago. We need to fix our documentation links if folks are continuing
>> to trip over it.
>>
>> HBase shell is documented here, off the main wiki page for hbase:
>> http://wiki.apache.org/hadoop/Hbase/Shell.
>>
>> To answer your question, there is nothing special to do adding new columns;
>> just create the column family and then add columns. As long as the columns
>> carry an existing column family prefix, hbase won't complain.
>>
>> To create a table 'x', with a column family 'x', do this in the shell:
>>
>> create 'x', 'x'
>>
>> Type 'help' in the shell for more info.
>>
>> 2) Also if I want to do it programmatically, HColumnDescriptor class only
>>
>>> lets you create column families with single columns. Is there a way to do
>>> it
>>> ?
>>>
>>>
>>>
>> See above. Yes, you only create the column family when defining your
>> schema. Thereafter you can add columns without modifying schema.
>>
>> Also If I am not wrong, we can store multiple values of each column in
>>
>>> each
>>> column family ?
>>>
>>>
>>>
>> Yes.
>>
>> Is there a way I can do a bulk import from a structured text file into
>>
>>> Hbase
>>> through the shell or any script already available
>>>
>>>
>> Usually folks write mapreduce jobs to load up hbase. If you want to write
>> a script to load hbase, see the aforementioned shell link. Has some notes
>> on how you to script the shell or on how to write a script in jruby.
>> Otherwise, there are python/jython examples here:
>> http://wiki.apache.org/hadoop/Hbase/Shell
>>
>> St.Ack
>>
>>
>
>
>
>
Re: Question of Column families in Hbase
Posted by Nishant Khurana <ni...@gmail.com>.
Hi,
Thanks for pointing me to the new link. A google search on Hbase Shell was
returning the old one at the top so was trying to use that.Sp from your
answers below, I think we always have to define column families in the
schema first and then alter table to add fully qualified column names to add
multiple columns to a column families.
I will try to use the scripts as mentioned there.
Thanks
On Sat, Nov 22, 2008 at 1:47 AM, Michael Stack <st...@duboce.net> wrote:
> Nishant Khurana wrote:
>
>> Hi,
>> I was looking at the API Docs and HQL shell to create tables for Hbase. I
>> couldn't figure out :
>> 1) How to create column families such that each column family contains
>> more
>> than one column through shell.
>>
>>
> Please tell us how you tripped over HQL. It was remove from hbase quite a
> while ago. We need to fix our documentation links if folks are continuing
> to trip over it.
>
> HBase shell is documented here, off the main wiki page for hbase:
> http://wiki.apache.org/hadoop/Hbase/Shell.
>
> To answer your question, there is nothing special to do adding new columns;
> just create the column family and then add columns. As long as the columns
> carry an existing column family prefix, hbase won't complain.
>
> To create a table 'x', with a column family 'x', do this in the shell:
>
> create 'x', 'x'
>
> Type 'help' in the shell for more info.
>
> 2) Also if I want to do it programmatically, HColumnDescriptor class only
>> lets you create column families with single columns. Is there a way to do
>> it
>> ?
>>
>>
> See above. Yes, you only create the column family when defining your
> schema. Thereafter you can add columns without modifying schema.
>
> Also If I am not wrong, we can store multiple values of each column in
>> each
>> column family ?
>>
>>
> Yes.
>
> Is there a way I can do a bulk import from a structured text file into
>> Hbase
>> through the shell or any script already available
>>
> Usually folks write mapreduce jobs to load up hbase. If you want to write
> a script to load hbase, see the aforementioned shell link. Has some notes
> on how you to script the shell or on how to write a script in jruby.
> Otherwise, there are python/jython examples here:
> http://wiki.apache.org/hadoop/Hbase/Shell
>
> St.Ack
>
--
Nishant Khurana
Candidate for Masters in Engineering (Dec 2009)
Computer and Information Science
School of Engineering and Applied Science
University of Pennsylvania
Re: Question of Column families in Hbase
Posted by Michael Stack <st...@duboce.net>.
Nishant Khurana wrote:
> Hi,
> I was looking at the API Docs and HQL shell to create tables for Hbase. I
> couldn't figure out :
> 1) How to create column families such that each column family contains more
> than one column through shell.
>
Please tell us how you tripped over HQL. It was remove from hbase quite
a while ago. We need to fix our documentation links if folks are
continuing to trip over it.
HBase shell is documented here, off the main wiki page for hbase:
http://wiki.apache.org/hadoop/Hbase/Shell.
To answer your question, there is nothing special to do adding new
columns; just create the column family and then add columns. As long as
the columns carry an existing column family prefix, hbase won't complain.
To create a table 'x', with a column family 'x', do this in the shell:
create 'x', 'x'
Type 'help' in the shell for more info.
> 2) Also if I want to do it programmatically, HColumnDescriptor class only
> lets you create column families with single columns. Is there a way to do it
> ?
>
See above. Yes, you only create the column family when defining your
schema. Thereafter you can add columns without modifying schema.
> Also If I am not wrong, we can store multiple values of each column in each
> column family ?
>
Yes.
> Is there a way I can do a bulk import from a structured text file into Hbase
> through the shell or any script already available
>
Usually folks write mapreduce jobs to load up hbase. If you want to
write a script to load hbase, see the aforementioned shell link. Has
some notes on how you to script the shell or on how to write a script in
jruby. Otherwise, there are python/jython examples here:
http://wiki.apache.org/hadoop/Hbase/Shell
St.Ack
Re: Question of Column families in Hbase
Posted by Tom Nichols <tm...@gmail.com>.
Once a family exists, you don't need to add columns. Just insert rows
using the family:column key. Think of it like a hashmap.
On Sat, Nov 22, 2008 at 12:27 AM, Nishant Khurana <ni...@gmail.com> wrote:
> Hi,
> I was looking at the API Docs and HQL shell to create tables for Hbase. I
> couldn't figure out :
> 1) How to create column families such that each column family contains more
> than one column through shell.
> 2) Also if I want to do it programmatically, HColumnDescriptor class only
> lets you create column families with single columns. Is there a way to do it
> ?