You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@hbase.apache.org by "Jim Kellerman (POWERSET)" <Ji...@microsoft.com> on 2008/12/16 04:49:19 UTC
RE: Performance Issue
---
Jim Kellerman, Powerset (Live Search, Microsoft Corporation)
> -----Original Message-----
> From: Dru Jensen [mailto:dru@brandaffinity.net]
> Sent: Monday, December 15, 2008 6:10 PM
> To: hbase-user@hadoop.apache.org
> Subject: Performance Issue
>
> I have an HBase table that stores calculated values for different
> dates and locations. This table only has 4 column families but with
> the combination of dates and locations generates 90K+ columns per
> row. The values stored are 4 characters long like "43.25". Reading
> the BigTable document, I decided to allow the number of columns to
> grow to large numbers.
>
> When I perform a get() on a single row, the results take 2 minutes to
> return. I was hoping for second response times. At first I thought
> this was related to the number of row in the table so I retried this
> with a table only having 1 row in it and the results were the same.
>
> Any ideas on how to fix this?
>
> Example:
>
> key=John Doe
> column:CA|2008-01-01 value=32.53
> column:CA|2008-01-02 value=32.54
> column:CA|2008-01-03 value=32.55
> column:CA|2008-01-04 value=32.56
> column:CA|2008-01-05 value=32.57
> column:CA|2008-01-06 value=32.58
> ...90,000+
>
> Thanks for you help!
> Dru Jensen
Re: Performance Issue
Posted by Jean-Daniel Cryans <jd...@apache.org>.
That's https://issues.apache.org/jira/browse/HBASE-867
Currently the only workaround is to undo the big columns. This really is a
critical issue for 0.20.0
J-D
On Mon, Dec 15, 2008 at 10:49 PM, Jim Kellerman (POWERSET) <
Jim.Kellerman@microsoft.com> wrote:
>
>
> ---
> Jim Kellerman, Powerset (Live Search, Microsoft Corporation)
>
>
> > -----Original Message-----
> > From: Dru Jensen [mailto:dru@brandaffinity.net]
> > Sent: Monday, December 15, 2008 6:10 PM
> > To: hbase-user@hadoop.apache.org
> > Subject: Performance Issue
> >
> > I have an HBase table that stores calculated values for different
> > dates and locations. This table only has 4 column families but with
> > the combination of dates and locations generates 90K+ columns per
> > row. The values stored are 4 characters long like "43.25". Reading
> > the BigTable document, I decided to allow the number of columns to
> > grow to large numbers.
> >
> > When I perform a get() on a single row, the results take 2 minutes to
> > return. I was hoping for second response times. At first I thought
> > this was related to the number of row in the table so I retried this
> > with a table only having 1 row in it and the results were the same.
> >
> > Any ideas on how to fix this?
> >
> > Example:
> >
> > key=John Doe
> > column:CA|2008-01-01 value=32.53
> > column:CA|2008-01-02 value=32.54
> > column:CA|2008-01-03 value=32.55
> > column:CA|2008-01-04 value=32.56
> > column:CA|2008-01-05 value=32.57
> > column:CA|2008-01-06 value=32.58
> > ...90,000+
> >
> > Thanks for you help!
> > Dru Jensen
>