[
https://issues.apache.org/jira/browse/HADOOP-1134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12483685
]
Doug Cutting commented on HADOOP-1134:
--------------------------------------
> Do you agree with the policy in principle?
I don't oppose it in principal, but might in practice. I think it should only
be pursued if we find that it offers significant performance improvements,
sufficient to offset increased complexity and consequent risk of introducing
checksum validation logic errors. We can benchmark this after this issue is
committed to find if it is warranted.
> How the over all interaction will look like in future will affect how we
> organize our code for this issue
True. So we should try to architect the protocol with future flexibility in
mind. But this is a private protocol, and there is still only a single client,
so revising the protocol in the future should also not be that difficult, so we
needn't prepare for every possibility now. I think the protocol I proposed
above, where read requests return something like <start, length, checksum,
data> records, would permit checksums to be re-generated at read-time or not,
no?
> Block level CRCs in HDFS
> ------------------------
>
> Key: HADOOP-1134
> URL: https://issues.apache.org/jira/browse/HADOOP-1134
> Project: Hadoop
> Issue Type: New Feature
> Components: dfs
> Reporter: Raghu Angadi
> Assigned To: Raghu Angadi
>
> Currently CRCs are handled at FileSystem level and are transparent to core
> HDFS. See recent improvement HADOOP-928 ( that can add checksums to a given
> filesystem ) regd more about it. Though this served us well there a few
> disadvantages :
> 1) This doubles namespace in HDFS ( or other filesystem implementations ). In
> many cases, it nearly doubles the number of blocks. Taking namenode out of
> CRCs would nearly double namespace performance both in terms of CPU and
> memory.
> 2) Since CRCs are transparent to HDFS, it can not actively detect corrupted
> blocks. With block level CRCs, Datanode can periodically verify the checksums
> and report corruptions to namnode such that name replicas can be created.
> We propose to have CRCs maintained for all HDFS data in much the same way as
> in GFS. I will update the jira with detailed requirements and design. This
> will include same guarantees provided by current implementation and will
> include a upgrade of current data.
>
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
