[ 
https://issues.apache.org/jira/browse/HBASE-2315?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12868989#action_12868989
 ] 

stack commented on HBASE-2315:
------------------------------

.bq I have started benchmarking hbase+bookkeeper with the Yahoo! benchmark, but 
I stumbled upon a problem. When I wrote this preliminary patch, I didn't have a 
nice way of creating a unique znode name and keeping it for fetching the last 
log of a region server, given the current interface of WAL. My understanding of 
the code is that the hdfs file where the is stored is passed to the 
reader/writer. I'm considering the current hdfs path as the znode path (or at 
least a part of it), but I would accept suggestions if anyone is willing to 
give me one.

The HDFS path should map to a znode path?  That'd work.  The WAL HDFS path has 
the regionserver name in it IIRC.  You probably don't need the full thing.  
Just from the regionserver name on down (should be shallow).

Let us know if you want us to change stuff in here around WAL.  You might want 
to wait a day or so till hbase-2437 goes in (refactoring of hlog).  Stuff 
should be easier to grok after that patch has had its way.

Is HBASE-2527 of any use to you?  It just went into trunk.

> BookKeeper for write-ahead logging
> ----------------------------------
>
>                 Key: HBASE-2315
>                 URL: https://issues.apache.org/jira/browse/HBASE-2315
>             Project: Hadoop HBase
>          Issue Type: New Feature
>          Components: regionserver
>            Reporter: Flavio Paiva Junqueira
>         Attachments: bookkeeperOverview.pdf, HBASE-2315.patch, 
> zookeeper-dev-bookkeeper.jar
>
>
> BookKeeper, a contrib of the ZooKeeper project, is a fault tolerant and high 
> throughput write-ahead logging service. This issue provides an implementation 
> of write-ahead logging for hbase using BookKeeper. Apart from expected 
> throughput improvements, BookKeeper also has stronger durability guarantees 
> compared to the implementation currently used by hbase.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.

Reply via email to