hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Michael Segel <michael_se...@hotmail.com>
Subject RE: Inserting data directly into HBase?
Date Tue, 01 Mar 2011 12:35:46 GMT


The short answer is yes.

I would have to guess that the reason you see examples of data first going to HDFS is that
they are using a map/reduce to insert the data. You don't have to do this.
You can make a connection from any "cloud aware" machine (Actually its just loading the proper
config data in to your java app. (And even that doesn't have to be java.)

The simplest thing is that you open a connection to HBase, instantiate an HTable object (target),
and then a Put object that you use to write to HBase.



> Date: Tue, 1 Mar 2011 20:28:58 +0900
> Subject: Inserting data directly into HBase?
> From: mp2893@gmail.com
> To: common-user@hadoop.apache.org; user@hbase.apache.org
> Hi,
> I am trying to crawl several thousands of rss feeds every 30 minutes.
> I thought I could use Hadoop and HBase as my platform.
> However, I am not familiar with the HBase architecture and was wondering if
> I could insert crawled news articles directly into HBase without first
> saving it into HDFS.
> I am asking this dumb question because all the HBase examples I saw in
> reference books are always starting with saving data to HDFS.
> And also, If I have 2 computers comprised of A for HDFS, and B for HBase,
> what happens when I insert data directly into HBase?
> Is the data stored in B automatically and a pointer is made to A?
> Or is the data stored in A and a pointer is made to itself?
> I really have no idea how HBase operates :(
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message