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 Thu, 31 May 2007 20:00:18 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-1134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12500467
] 

Doug Cutting commented on HADOOP-1134:
--------------------------------------

Some comments on the design doc:

1. .meta seems like a more obvious extension than .mtd for metadata.

2. Re: "common header": should packets be length-prefixed?  That might make processing easier.

3. Re: "DATA_CHUNK": Unless I misunderstand, this only supports a single checksum per chunk
and a single chunk per packet.  So, if we want to support the current 512-byte checksums,
this requires a packet every 512 bytes, which seems small to me.  I'd prefer something like:

<header, OP=WRITE, start, length, bytesPerChecksum, <data, checksum>*>

So, if length<=bytesPerChecksum, there's just a single checksum, as in your proposal, but
if length > bytesPerChecksum then there are multiple checksums in the data.


> 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
>            Assignee: Raghu Angadi
>         Attachments: bc-no-upgrade-05302007.patch, DfsBlockCrcDesign-05305007.htm
>
>
> 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.


Mime
View raw message