You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@hbase.apache.org by hongbin ma <ma...@apache.org> on 2015/12/28 07:19:27 UTC
performance difference between observer and endpoint
hi experts,
we're working on using observer/endpoint to visit hbase rows. During our
experiments we found that for *small interval scans* (where the scan range
is within a single region, and the scan range contains only tens/hundreds
of rows), observer is times faster than endpoint. Our question is, for *small
interval scans, *is observer naturally a better choice than endpoint?
I do know how to use these two coprocessors, however I'm not quite clear
about the fundamental difference of them that might lead to performance
gap. Any documentations or insights are greatly appreciated.
--
Regards,
*Bin Mahone | 马洪宾*
Apache Kylin: http://kylin.io
Github: https://github.com/binmahone
Re: performance difference between observer and endpoint
Posted by Anoop John <an...@gmail.com>.
And which exact version you are using?
-Anoop-
On Tue, Dec 29, 2015 at 12:50 PM, Anoop John <an...@gmail.com> wrote:
> Mind sharing some more details like what exactly u are trying to do with
> these scans? How is this observers look like and the EPs. What abt more
> bigger range scans? Any perf diff?
>
> -Anoop-
>
> On Mon, Dec 28, 2015 at 11:49 AM, hongbin ma <ma...@apache.org> wrote:
>
>> hi experts,
>>
>> we're working on using observer/endpoint to visit hbase rows. During our
>> experiments we found that for *small interval scans* (where the scan range
>> is within a single region, and the scan range contains only tens/hundreds
>> of rows), observer is times faster than endpoint. Our question is, for
>> *small
>> interval scans, *is observer naturally a better choice than endpoint?
>>
>> I do know how to use these two coprocessors, however I'm not quite clear
>> about the fundamental difference of them that might lead to performance
>> gap. Any documentations or insights are greatly appreciated.
>>
>> --
>> Regards,
>>
>> *Bin Mahone | 马洪宾*
>> Apache Kylin: http://kylin.io
>> Github: https://github.com/binmahone
>>
>
>
Re: performance difference between observer and endpoint
Posted by Anoop John <an...@gmail.com>.
Mind sharing some more details like what exactly u are trying to do with
these scans? How is this observers look like and the EPs. What abt more
bigger range scans? Any perf diff?
-Anoop-
On Mon, Dec 28, 2015 at 11:49 AM, hongbin ma <ma...@apache.org> wrote:
> hi experts,
>
> we're working on using observer/endpoint to visit hbase rows. During our
> experiments we found that for *small interval scans* (where the scan range
> is within a single region, and the scan range contains only tens/hundreds
> of rows), observer is times faster than endpoint. Our question is, for
> *small
> interval scans, *is observer naturally a better choice than endpoint?
>
> I do know how to use these two coprocessors, however I'm not quite clear
> about the fundamental difference of them that might lead to performance
> gap. Any documentations or insights are greatly appreciated.
>
> --
> Regards,
>
> *Bin Mahone | 马洪宾*
> Apache Kylin: http://kylin.io
> Github: https://github.com/binmahone
>