hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Doug Cutting (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-1134) Block level CRCs in HDFS
Date Tue, 27 Mar 2007 18:43:32 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-1134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12484538

Doug Cutting commented on HADOOP-1134:

> 64KB is only for new blocks.

I thought we'd benchmark before we changed any parameters, no?  I'd vote to keep this at 512
until we benchmark.

> I personally prefer datanode verification (and general ownership of CRCs). I don't think
it adds noticeable latency.

This sounds like another change we should benchmark before we make.  This patch should confine
itself to a single issue: removing the checksum data from the DFS namespace.  Subsequently
we should determine the performance impacts of double-checksumming (on both datanode &
client), larger checksum windows, etc., before committing such changes.

Changing the storage of checksum data should alone improve performance, since fewer datanode
and namenode accesses are required.  It will be useful to see how much this change alone improves
things before we start making other optimizations.

> 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.

View raw message