[
https://issues.apache.org/jira/browse/HDFS-8430?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15080487#comment-15080487
]
Tsz Wo Nicholas Sze commented on HDFS-8430:
-------------------------------------------
[~drankye], I think it is a good start for the first implementation. We may
improve it later on. Some ideas:
# Instead of sending all CRCs to the client, send all CRCs to one of the
datanode in a block group. The datanode computes the block MD5s and returns
them to the client. Then, the computation becomes distributed.
# We may consider changing the checksum algorithm for replicated files
(although it is incompatible with the old clusters)
## Use CRC64 (or some other linear code) for block checksum instead of MD5.
The datanode may compute cell CRC64s and then send them to a client (or a
datanode). We may combine the cell CRC64s to obtain the block CRC64 since the
code is linear. Since datanodes send cell checksums instead of data checksums,
the network overhead becomes ignorable.
## Or simply compute cell checksums for replicated files instead of block
checksums.
> Erasure coding: update DFSClient.getFileChecksum() logic for stripe files
> -------------------------------------------------------------------------
>
> Key: HDFS-8430
> URL: https://issues.apache.org/jira/browse/HDFS-8430
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Affects Versions: HDFS-7285
> Reporter: Walter Su
> Assignee: Kai Zheng
> Attachments: HDFS-8430-poc1.patch
>
>
> HADOOP-3981 introduces a distributed file checksum algorithm. It's designed
> for replicated block.
> {{DFSClient.getFileChecksum()}} need some updates, so it can work for striped
> block group.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
