cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "shawndu (JIRA)" <j...@apache.org>
Subject [jira] [Created] (CASSANDRA-4418) Add new file Column family Type
Date Fri, 06 Jul 2012 10:05:34 GMT
shawndu created CASSANDRA-4418:
----------------------------------

             Summary: Add new file Column family Type
                 Key: CASSANDRA-4418
                 URL: https://issues.apache.org/jira/browse/CASSANDRA-4418
             Project: Cassandra
          Issue Type: Improvement
          Components: Core
    Affects Versions: 1.1.2
         Environment: All
            Reporter: shawndu
             Fix For: 1.1.2


Now I think cassandra's main problems is the compaction.
Our use case is:
we insert many big column value data in cassandra. column value maybe more than 10k.for cassandra
write is quiet fast. it will cause cassandra flush memtbale data into sstable.but these data
may be deleted soon. so issues come: compaction.
I don't know the root cause. but I know the result. in this case, compaction will make server
become slow and also very high CPU and at a dangerous state. for we have other much important
process running.

I think we need a new File Column Family type. It is:
a Column value is not a byte array but a file name. the real value is in the file.
there are some benefits:
0 when column is inserted, just create the file.
1 when column value is update. just replace the file.
2 when column is deleted, just delete the file.
3 this will save much time when compaction.

I don't know it is a good, but in my view, it is simple and can real fix the compaction problems.




--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Mime
View raw message