[
https://issues.apache.org/jira/browse/HDFS-7662?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14296406#comment-14296406
]
Kai Zheng commented on HDFS-7662:
---------------------------------
Hi [~zhz],
Thanks for your review.
bq.If that's the case, is it better to keep rawcoder under hadoop common, and
move coder to hdfs?
Yes it's the case. *coder* works with blocks and block groups. I'm thinking
that if the whole erasure codec framework can be out with some certain
abstraction getting rid of HDFS specific details it's worth the effort. I saw
other codec like frameworks like compression and encryption were placed in
Common side already. While I'm breaking down the codes in HDFS-7337 and sorting
out the details, I thought it's doable though the left patches are to be out.
bq.Maybe change coder to blockcoder
Good idea, will do so. Thanks !
bq.ECBlock class has no pointer to identify an HDFS block. How should the
client / DN specify what HDFS blocks to encode / decode?
Yes it's an abstraction and doesn't involve into HDFS details. HDFS block can
extend it to do HDFS specific mapping logic. I will try to accelerates this to
come up the left patches so you can see the whole picture.
bq.Could you post an example how to use the callback class under this JIRA?
Good idea, I will provide an example but am not so sure in this JIRA.
> Erasure Coder API for encoding and decoding of block group
> ----------------------------------------------------------
>
> Key: HDFS-7662
> URL: https://issues.apache.org/jira/browse/HDFS-7662
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Reporter: Kai Zheng
> Assignee: Kai Zheng
> Attachments: HDFS-7662-v1.patch
>
>
> This is to define ErasureCoder API for encoding and decoding of BlockGroup.
> Given a BlockGroup, ErasureCoder extracts data chunks from the blocks and
> leverages RawErasureCoder defined in HDFS-7353 to perform concrete encoding
> or decoding.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
