[
https://issues.apache.org/jira/browse/HADOOP-11647?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14357998#comment-14357998
]
Kai Zheng commented on HADOOP-11647:
------------------------------------
Thanks [~umamaheswararao] for the review and comments !
bq.Seems like you did not load this item from configuration.
I planned to do it in {{ErasureCodec}} layer in HADOOP-11645. But you're right
it should be done here now and I will.
bq.why do we name this API as performDecoding ?
Good question. The *perform* word is too strong and really questionable. I will
think about better name.
bq.useXorWhenPossible: this optimization does not apply for encoding case?
Yes, it's only for recovering. In recovering, most time there would be only ONE
erased block to be recovered, in this case the simpler XOR procedure will be
much faster. In encoding, we will rarely have schema of only ONE parity block
to be generated.
bq.May be you can release then and there when encoding/decoding done
respectively?
You're right, it should.
bq.change to Reed-Solomon
Sure.
Will update the patch today accordingly. Thanks !
> Reed-Solomon ErasureCoder
> -------------------------
>
> Key: HADOOP-11647
> URL: https://issues.apache.org/jira/browse/HADOOP-11647
> Project: Hadoop Common
> Issue Type: Sub-task
> Reporter: Kai Zheng
> Assignee: Kai Zheng
> Attachments: HADOOP-11647-v2.patch, HDFS-7664-v1.patch
>
>
> This is to implement Reed-Solomon ErasureCoder using the API defined in
> HADOOP-11646. It supports to plugin via configuration for concrete
> RawErasureCoder, using either JRSErasureCoder added in HDFS-7418 or
> IsaRSErasureCoder added in HDFS-7338.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
