hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ted Yu <yuzhih...@gmail.com>
Subject Re: Coprocessor Aggregation supposed to be ~20x slower than Scans?
Date Mon, 14 May 2012 19:55:37 GMT
Anil:
I think the performance was related to your custom filter.

Please tell us more about the filter next time.

Thanks

On Mon, May 14, 2012 at 12:31 PM, anil gupta <anilgupta84@gmail.com> wrote:

> HI Stack,
>
> I'll look into Gary Helming post and try to do profiling of coprocessor and
> share the results.
>
> Thanks,
> Anil Gupta
>
> On Mon, May 14, 2012 at 12:08 PM, Stack <stack@duboce.net> wrote:
>
> > On Mon, May 14, 2012 at 12:02 PM, anil gupta <anilgupta84@gmail.com>
> > wrote:
> > > I loaded around 70 thousand 1-2KB records in HBase. For scans, with my
> > > custom filter i am able to get 97 rows in 500 milliseconds and for
> doing
> > > sum, max, min(in built aggregations of HBase) on the same custom filter
> > its
> > > taking 11000 milliseconds. Does this mean that coprocessors aggregation
> > is
> > > supposed to be around ~20x slower than scans? Am i missing any trick
> over
> > > here?
> > >
> >
> > That seems like a high tax to pay for running CPs.  Can you dig in on
> > where the time is being spent?  (See another recent note on this list
> > or on dev where Gary Helmling talks about how he did basic profiling
> > of CPs).
> > St.Ack
> >
>
>
>
> --
> Thanks & Regards,
> Anil Gupta
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message