[ 
https://issues.apache.org/jira/browse/HBASE-5074?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13201614#comment-13201614
 ] 

Phabricator commented on HBASE-5074:
------------------------------------

dhruba has commented on the revision "[jira] [HBASE-5074] Support checksums in 
HBase block cache".

  Todd: I agree with you. It is messy that the HFileSystem interface is leaking 
out to the unit tests. Instead, inside HFile, I can do something like this when 
a Reader is created:

  if (!fs instanceof HFileSystem) {
    fs = new HFileSystem(fs);
  }

  what this means is that users of HFile that already passes in a HFileSystem 
will get the new behaviour while. HReginServer anyways voluntarily creates 
HFileSystem before invoking HFile, so it work.

  I did not do this earlier because I thought that 'using reflection' is 
costly, but on second thoughts the cost is not much because it will be done 
only once when a new reader is created for the first time. what do you think?


REVISION DETAIL
  https://reviews.facebook.net/D1521

                
> support checksums in HBase block cache
> --------------------------------------
>
>                 Key: HBASE-5074
>                 URL: https://issues.apache.org/jira/browse/HBASE-5074
>             Project: HBase
>          Issue Type: Improvement
>          Components: regionserver
>            Reporter: dhruba borthakur
>            Assignee: dhruba borthakur
>         Attachments: D1521.1.patch, D1521.1.patch, D1521.2.patch, 
> D1521.2.patch, D1521.3.patch, D1521.3.patch
>
>
> The current implementation of HDFS stores the data in one block file and the 
> metadata(checksum) in another block file. This means that every read into the 
> HBase block cache actually consumes two disk iops, one to the datafile and 
> one to the checksum file. This is a major problem for scaling HBase, because 
> HBase is usually bottlenecked on the number of random disk iops that the 
> storage-hardware offers.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: 
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Reply via email to