You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@hbase.apache.org by Patrick Räuber <Pa...@gmx.de> on 2011/11/22 16:33:26 UTC

get performance

Hi everybody !

I'm doing a get (from shell) on a row containing 10 - 20 counter values (8 byte longs) that are currently incremented which takes about 700 - 800ms.

=> get 'table', 'row'

The same gets for older rows return in about 20 - 30ms (same number of values..).

=> get 'table', 'row'

Is this related to row locking ? 

What further confuses me : 

Doing the get on the 'slow' row when explicitly specifying columns to retrieve makes it fast again. 

=> get 'table', 'row', {COLUMNS => [...]}

Could someone enlight me ? ;)

Thx ! 

Patrick

-- 
NEU: FreePhone - 0ct/min Handyspartarif mit Geld-zurück-Garantie!		
Jetzt informieren: http://www.gmx.net/de/go/freephone

Re: get performance

Posted by Doug Meil <do...@explorysmedical.com>.
What could be happening is this...

http://hbase.apache.org/book.html#client

... the first time a JVM starts it has to look up region metadata and then
cache it in the JVM.  The first get might be experiencing that.






On 11/22/11 10:33 AM, "Patrick Räuber" <Pa...@gmx.de> wrote:

>Hi everybody !
>
>I'm doing a get (from shell) on a row containing 10 - 20 counter values
>(8 byte longs) that are currently incremented which takes about 700 -
>800ms.
>
>=> get 'table', 'row'
>
>The same gets for older rows return in about 20 - 30ms (same number of
>values..).
>
>=> get 'table', 'row'
>
>Is this related to row locking ?
>
>What further confuses me :
>
>Doing the get on the 'slow' row when explicitly specifying columns to
>retrieve makes it fast again.
>
>=> get 'table', 'row', {COLUMNS => [...]}
>
>Could someone enlight me ? ;)
>
>Thx ! 
>
>Patrick
>
>-- 
>NEU: FreePhone - 0ct/min Handyspartarif mit Geld-zurück-Garantie!		
>Jetzt informieren: http://www.gmx.net/de/go/freephone
>