hbase-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Aditya Kumar <adityakumar...@yahoo.com>
Subject Re: Hbase -question
Date Tue, 15 Nov 2011 07:00:31 GMT
Thanks for the reply and explanation.

Can you explain me the following:
HFile is having the raw data blocks and also the indices to find the the blocks  about the
i.e, It has the raw data blocks and indices of these blocks?

will it not be heavy to have multiple copies?
I was of the understanding that : Hfile will have only the filename with the indices of the

can you please clarify me more in detail why we need to have the raw data also?

From: Stack <stack@duboce.net>
To: dev@hbase.apache.org; Aditya Kumar <adityakumar345@yahoo.com>
Sent: Monday, November 14, 2011 10:37 PM
Subject: Re: Hbase -question

On Mon, Nov 14, 2011 at 10:25 PM, Aditya Kumar <adityakumar345@yahoo.com> wrote:
> can any one please clarify me :
> what is HLog,storefile.Hfile and store?

HLog is the name of our WAL implementation (see in code the wal
package, see how the WAL implementation is done in the HLog class).

HFile is about raw data blocks and indices to find those blocks as
well as metadata about the file.  Ideally it knows nothing about the
upper hbase tiers.  StoreFile then is the glue between hbase and its
raw storage HFile.

Store is the implementation of a Column Family (The Store class should
have been called ColumnFamily)

Sorry our naming is a little less clear than it should be.

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message