Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
HBase >> mail # user >> Single table Vs Multiple table. Need inputs !!!


Copy link to this message
-
RE: Single table Vs Multiple table. Need inputs !!!
Hi
>Create separate table for each report type and manage queries using  joins
There is no join support in HBase. You need to handle it in your client code.

See whether you can implement your requirement using one table approach.

-Anoop-
________________________________________
From: AnandaVelMurugan Chandra Mohan [[EMAIL PROTECTED]]
Sent: Wednesday, June 13, 2012 3:07 PM
To: [EMAIL PROTECTED]
Subject: Single table Vs Multiple table. Need inputs !!!

Hi,

I am planning to use HBase DB for storing multiple reports of a component.
Each report has a unique set of metadata.

My query requirement is like this. I need to query and find out number of
reports of each component. Then I would have to filter those reports based
on some random metadata.

Initially I decided to use one table with multiple column families to
handle each report type. But HBase documentation instructs us to keep the
column families minimal. Also if a new report type gets added in future, I
will not be able to add new column family.

So I have two alternatives,
   - Have one column family for report metadata and fill the values in
   column based on report type. This will create sparse rows which I can
   accommodate.
   - Create separate table for each report type and manage queries using
   joins.

Any one have any better alternatives? Anyone see any performance
bottlenecks in these two alternatives. Please help me in deciding a durable
solutions. Thanks.
--
Regards,
Anand