You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@hbase.apache.org by Ski Gh3 <sk...@gmail.com> on 2008/10/20 23:49:56 UTC
log replication in hbase
Just want to make sure my understanding of the log is correct:
each regionserver has one log for all the regions it is serving.
this log is the write-ahead log that is stored on hadoop dfs and always
written before a data is written to memcache.
is this correct?
I read about a redo log from the mailing list, which says each regionserver
maintains a redo log locally.
is this the wal log or what is the purpose of this redo log?
Thanks!
RE: log replication in hbase
Posted by "Jim Kellerman (POWERSET)" <Ji...@microsoft.com>.
> -----Original Message-----
> From: Ski Gh3 [mailto:skigh3@gmail.com]
> Sent: Tuesday, October 21, 2008 10:43 AM
> To: hbase-user@hadoop.apache.org
> Subject: Re: log replication in hbase
>
> The last "yes" is contradictary with the previous one...
> So I take this redo log is the WAL log?
Yes.
> It is written to HDFS (replicated three times) rather than stored on the
> regionserver itself locally?
Yes.
> Thanks!
> On Mon, Oct 20, 2008 at 9:33 PM, stack <st...@duboce.net> wrote:
>
> > Ski Gh3 wrote:
> >
> >> Just want to make sure my understanding of the log is correct:
> >>
> >> each regionserver has one log for all the regions it is serving.
> >>
> >>
> >
> > Yes.
> >
> > this log is the write-ahead log that is stored on hadoop dfs and
> always
> >> written before a data is written to memcache.
> >> is this correct?
> >>
> >>
> >
> > Yes.
> >
> > I read about a redo log from the mailing list, which says each
> >> regionserver
> >> maintains a redo log locally.
> >> is this the wal log or what is the purpose of this redo log?
> >>
> >>
> > Yes.
> >
> > Sorry for any confusion. We should be try and be more strict with our
> > nomenclature.
> >
> > St.Ack
> >
> >
Re: log replication in hbase
Posted by Ski Gh3 <sk...@gmail.com>.
The last "yes" is contradictary with the previous one...
So I take this redo log is the WAL log?
It is written to HDFS (replicated three times) rather than stored on the
regionserver itself locally?
Thanks!
On Mon, Oct 20, 2008 at 9:33 PM, stack <st...@duboce.net> wrote:
> Ski Gh3 wrote:
>
>> Just want to make sure my understanding of the log is correct:
>>
>> each regionserver has one log for all the regions it is serving.
>>
>>
>
> Yes.
>
> this log is the write-ahead log that is stored on hadoop dfs and always
>> written before a data is written to memcache.
>> is this correct?
>>
>>
>
> Yes.
>
> I read about a redo log from the mailing list, which says each
>> regionserver
>> maintains a redo log locally.
>> is this the wal log or what is the purpose of this redo log?
>>
>>
> Yes.
>
> Sorry for any confusion. We should be try and be more strict with our
> nomenclature.
>
> St.Ack
>
>
Re: log replication in hbase
Posted by stack <st...@duboce.net>.
Ski Gh3 wrote:
> Just want to make sure my understanding of the log is correct:
>
> each regionserver has one log for all the regions it is serving.
>
Yes.
> this log is the write-ahead log that is stored on hadoop dfs and always
> written before a data is written to memcache.
> is this correct?
>
Yes.
> I read about a redo log from the mailing list, which says each regionserver
> maintains a redo log locally.
> is this the wal log or what is the purpose of this redo log?
>
Yes.
Sorry for any confusion. We should be try and be more strict with our
nomenclature.
St.Ack